Você está na página 1de 237

Anders Rooma Nielsen

Ph.D. Tesis
March 2012
Combustion of large solid fuels
in cement rotary kilns


Anders Rooma Nielsen
PhD Thesis
FLSmidth A/S
Research & Development
1
Combustion of large solid fuels
in cement rotary kilns
March 2012
2
Copyright: Anders Rooma Nielsen
March 2012
Address: Centre of Combustion and Harmful Emission Control
Department of Chemical and
Biochemical Engineering
Technical University of Denmark
Sltofts Plads, Building 229
DK-2800 Kgs. Lyngby
Denmark
Phone: +45 4525 2800
Fax: +45 4525 4588
Web: www.chec.dtu.dk
Print: J&R Frydenberg A/S
Kbenhavn
March 2012
ISBN: 978-87-92481-66-5
i
Preface
This PhD project is part of an effort to strengthen the scientific based knowledge within
alternative fuel combustion technology which is becoming a more integrated part of cement
plants. The new knowledge obtained from the PhD project contributes to maintain FLSmidths
position as a technological leader in this field.
The project has been made in cooperation with the CHEC research group at the Department of
Chemical and Biochemical Engineering at the Technical University of Denmark (DTU) and the
Research and Development Department of FLSmidth A/S. The project has been partly sponsored
by The Danish National Advanced Technology Foundation.
I would like to thank my supervisors Morten Boberg Larsen (FLSmidth), Kim Dam-Johansen
(DTU) and Peter Glarborg (DTU) for all the time and patience they have invested in the
supervision of this project. Also special thanks to the people at FLSmidths research center
Dania, the CHEC technicians group and the workshop at DTU for their support regarding the
construction of a high-temperature rotary drum for combustion experiments as well as other
matters. Thanks to all the other people at FLSmidth and CHEC, not mentioned here, who have
also been involved in this project.
Finally, I would also like to thank the other PhD-students at CHEC, particularly those involved
in the co-operation with FLSmidth, for providing a pleasant working atmosphere, for many
fruitful discussions and for the social activities we have had.
Anders Rooma Nielsen
Valby, March 2012
3
ii
4
iii
Abstract
The cement industry has a significant interest in replacing fossil fuels with alternative fuels in
order to minimize production costs and reduce CO
2
emissions. These new alternative fuels are in
particular solid fuels such as refuse derived fuel (RDF), tire-derived fuel (TDF), meat and bone
meal (MBM), waste wood, sewage sludge, paper and plastics. The alternative fuel share of the
total energy varies significantly from region to region, but the general trend is towards increased
alternative fuel utilization. Solid alternative fuels typically have physical and chemical properties
that differ from traditional solid fossil fuels. This creates a need for new combustion equipment
or modification of existing kiln systems, because alternative fuels may influence process stability
and product quality. Process stability is mainly influenced by exposing the raw material bed in
the rotary kiln to reducing conditions, which increases the tendency for deposit formations in the
rotary kiln material inlet end, kiln riser duct and lower cyclone stages. Clinker quality may also
be affected by minor compounds from the fuel ashes or from unburned carbon leaving the rotary
kiln with the clinker.
This thesis provides an insight into the utilization of solid alternative fuels in the material inlet
end of rotary kilns. This position is interesting because it allows utilization of large fuel particles,
thereby eliminating the need for an expensive shredding of the fuels. The challenge, however, is
that the solid fuels will be mixed into the cement raw materials, which is likely to affect process
stability and clinker quality, as described above.
The mixing of fuels and raw materials was studied experimentally in a pilot-scale rotary drum
and was found to be a fast process, reaching steady state within few drum revolutions. Thus, heat
transfer by conduction from the cement raw materials to the fuel particles is a major heat transfer
mechanism rather than convection or radiation from the freeboard gas above the material bed.
Consequently, the temperature of the cement raw materials becomes a factor of great importance
for heating the fuel particles. Combustion of different alternative fuels has been investigated
experimentally in a pilot-scale rotary furnace under conditions similar to those in the material
inlet end of cement rotary kilns. The main focus was on tire rubber and pine wood which are
relevant fuels in this context. Heating, drying and devolatilization of alternative fuels are fast
processes that primarily depend on heat transfer and fuel particle size. Devolatilization of a large
wood or tire particle with a thickness of 20 mm at 900C is for example around 2 minutes. By
contrast, char oxidation is a slow process which may greatly reduce the amounts of solid fuels to
be utilized in the material inlet end of rotary kilns due to the limited residence time. Several
parameters control the rate of char oxidation: a) bulk oxygen concentration, b) mass transfer rate
of oxygen to char particles, c) conversion pathway, d) bed material fill degree and e) char
particle size and shape. Parameters such as temperature and rotational speed only have a minor
influence on char oxidation subject to the conditions in cement rotary kilns. Models for
devolatilization and char combustion of tire rubber and pine wood have been developed and
compared with experimental results. The models may be modified to qualitatively predict
conversion times in industrial rotary kilns. This will, however, require further model
development and preferably validation against full-scale data.
Sulfur release from cement raw materials during alternative fuel combustion have been
investigated both experimentally and with thermodynamical equilibrium calculations. Known
effects of temperature and gas atmosphere on the decomposition of sulfates in the raw materials
5
iv
were confirmed. In addition, new knowledge was obtained regarding the effects of alternative
fuel types and fuel particle sizes on sulfur release: Particularly tire rubber led to a high sulfur
release. For all tested alternative fuels, the sulfur release from raw materials was observed to
primarily take place during fuel devolatilization where the rapid formation of reducing agents
such as CO led to high sulfur release from the raw materials. It was found that the overall sulfur
release could be reduced by using larger fuel particles due to a slower devolatilization rate and a
reduced tendency for local reducing conditions.
6
v
Resume (in Danish)
Cementindustrien har en stor interesse i at erstatte fossile brndsler med alternative brndsler,
med henblik p at minimere omkostninger og reducere CO
2
-emissioner. Disse alternative
brndsler er primrt faste brndsler som affaldsbaserede brndsler (RDF), brugte dk (TDF),
kd- og benmel (MBM), traffald, kloakslam, papir, pap og plastic. Alternativ brndsels
andelen af den totale energi varierer meget fra region til region, men generelt er der en tendens til
stigende anvendelse af alternative brndsler. Faste alternative brndsler har typisk anderledes
fysiske og kemiske egenskaber end de traditionelle faste fossile brndsler. Dette skaber et behov
for nyt forbrndingsudstyr eller modifikationer p eksisterende ovnsystemer fordi, at alternative
brndsler kan pvirke processtabiliteten og produktkvaliteten. Processtabiliteten pvirkes
hovedsageligt ved at eksponere rmaterialefyldningen i roterovnen for reducerende atmosfrer,
hvilket forger tendensen for aflejringer i materialeindlbet p roterovnen, stigrret og i de nedre
cyklontrin. Klinkerkvaliteten kan ogs pvirkes af grundstoffer i brndselsasken eller af
uforbrndt kulstof, der transporteres ud af roterovnen med klinkerne.
Denne afhandling giver indsigt i anvendelsen af faste alternative brndsler i materialeindlbet p
roterovne. Dette indfyringspunkt er interessant fordi, at det tillader anvendelse af store
brndselspartikler, hvormed behovet for en omkostningstung neddeling af brndslerne kan
undgs. Udfordringen er dog, at de faste brndsler vil blive blandet med cementrmaterialerne,
hvilket kan pvirke processtabiliteten og klinkerkvaliteten, som beskrevet ovenfor.
Opblandingen af brndsler og rmaterialer er blevet undersgt eksperimentelt i en pilotskala
rotertromle, og viste sig at finde sted hurtigt, med opnelse af fuld opblanding i lbet af f
tromleomdrejninger. Dette medfrer, at varmeovergang ved konduktion fra
cementrmaterialerne til brndselspartiklerne m forventes at vre en dominerende
varmeovergangsmekanisme i forhold til konvektion eller strling fra gassen over
rmaterialefyldningen. Temperaturen af cementrmaterialerne bliver derfor af stor betydning for
opvarmningen af brndselspartiklerne. Forbrnding af forskellige alternative brndsler er blevet
undersgt eksperimentelt i en pilotskala roterovn under procesforhold svarende til dem i
materialeindlbet p industrielle roterovne. Fokus var p dkgummi og fyrretr, som er
relevante brndsler i denne kontekst. Opvarmning, trring og devolatilisation af alternative
brndsler er hurtige processer, som primrt afhnger af opvarmningshastighed og
partikelstrrelse. Devolatilisation af en stor fyrretrspartikel med en tykkelse p 20 mm ved
900C tager for eksempel omkring to minutter. Koksoxidation er derimod en langsom proces,
hvilket kan reducere mngderne af faste brndsler i materialeindlbet p roterovne betydeligt,
p grund af den begrnsede opholdstid. Adskillige parametre pvirker hastigheden af
koksoxidationen: a) iltkoncentrationen, b) masseovergang af ilt til kokspartiklen, c)
omdannelsesvejen, d) fyldningsgraden i roterovnen og e) kokspartiklens strrelse og form.
Parametre som temperatur og roterhastighed har kun en mindre indflydelse p koksoxidationen
under de forhold, som er til stede i cement-roterovne. Modeller for devolatilisation og
koksoxidation af dkgummi og fyrretr er blevet udviklet og sammenlignet med
eksperimentelle resultater. Modellerne er blevet modificeret til kvalitativt at kunne estimere
7
vi
udbrndingstider i industrielle roterovne. Yderligere modeludvikling og eventuelt
sammenligning med fuldskala-data er dog pkrvet.
Svovlfrigivelse fra cementrmaterialerne under forbrndingen af alternative brndsler er blevet
studeret bde eksperimentelt og med termodynamiske ligevgtsberegninger. Kendte effekter af
temperatur og gas atmosfre p dekomponeringen af sulfater i rmaterialerne er blevet
bekrftet. Derudover er der fremkommet ny viden om effekten af forskellige alternative
brndsler og brndselspartikelstrrelsen p svovlfrigivelsen: Isr forbrnding af dkgummi
medfrer hj svovlfrigivelse fra rmaterialerne. For alle testede brndsler fandt svovlfrigivelsen
fra rmaterialerne primrt sted under devolatilisationen, hvor hurtig frigivelse af
reduktionsmidler som CO medfrte hj svovlfrigivelse fra rmaterialerne. Svovlfrigivelsen
kunne reduceres ved at benytte strre brndselspartikler p grund af en langsommere
devolatilisation og dermed en reduceret tendens til lokale reducerende forhold i fyldningen.
8
vii
Table of Contents
1 Introduction........................................................................................................................... 1
1.1 Background ...................................................................................................................... 1
1.2 Structure ........................................................................................................................... 2
1.3 Objective .......................................................................................................................... 3
1.4 References ........................................................................................................................ 4
2 Cement Chemistry and Production..................................................................................... 5
2.1 Cement Chemistry and Processing................................................................................... 5
2.1.1 History.................................................................................................................................................5
2.1.2 Raw materials ......................................................................................................................................5
2.1.3 Preparation of raw materials................................................................................................................6
2.1.4 Preheating of raw meal........................................................................................................................7
2.1.5 Calcination ..........................................................................................................................................7
2.1.6 Clinker reactions..................................................................................................................................8
2.1.7 Clinker Cooling .................................................................................................................................10
2.1.8 Clinker grinding and storage .............................................................................................................10
2.2 The kiln system.............................................................................................................. 11
2.2.1 Rotary kiln.........................................................................................................................................11
2.2.2 Design of modern rotary kilns for ILC systems ................................................................................13
2.2.3 Preheater and calciner systems..........................................................................................................14
2.2.4 In-Line Calciner (ILC) system..........................................................................................................15
2.3 Overview of process data ............................................................................................... 17
2.4 Conclusions for chapter 2............................................................................................... 18
2.5 References ...................................................................................................................... 19
3 Alternative fuels in cement production............................................................................... 21
3.1 Requirements for alternative fuels in cement production .............................................. 21
3.2 Types and quantities of alternative fuels........................................................................ 23
3.2.1 Energy share by alternative fuels.......................................................................................................23
3.2.2 Types and quantities of alternative fuels ...........................................................................................24
3.2.3 Fuel analysis......................................................................................................................................24
3.3 Technologies for alternative fuels combustion .............................................................. 27
3.3.1 HOTDISC..........................................................................................................................................27
3.3.2 The Pre-Combustion Chamber (PCC) ...............................................................................................28
3.3.3 Gasifiers ............................................................................................................................................29
3.3.4 Kiln material inlet firing and mid-kiln firing.....................................................................................29
9
viii
3.4 Effect of alternative fuels on process stability and product quality............................... 30
3.4.1 Cement chemistry..............................................................................................................................30
3.4.2 Emissions ..........................................................................................................................................32
3.4.3 Local reducing conditions .................................................................................................................34
3.4.4 Alkali, sulfur and chlorine: Circulation in the kiln system................................................................36
3.5 Conclusion...................................................................................................................... 38
3.6 References ...................................................................................................................... 39
4 Mixing of large and small particles in rotary kilns.......................................................... 45
4.1 Introduction.................................................................................................................... 45
4.2 Literature study on mixing and segregation in rotary kilns and drums.......................... 48
4.3 Experimental .................................................................................................................. 50
4.3.1 Experimental set-up...........................................................................................................................50
4.3.2 Bed materials and fuel characteristics ...............................................................................................51
4.3.3 Experimental procedure.....................................................................................................................52
4.3.4 Assumptions and uncertainties ..........................................................................................................53
4.4 Results and discussions .................................................................................................. 54
4.4.1 General observations .........................................................................................................................54
4.4.2 Effect of fuel particle size and shape.................................................................................................55
4.4.3 Effect of bed fill degree.....................................................................................................................58
4.4.4 Effect of bed rotational speed............................................................................................................59
4.4.5 Effect of fuel density .........................................................................................................................60
4.4.6 Effect of tumblers..............................................................................................................................61
4.5 Practical implications ..................................................................................................... 63
4.6 Conclusions for Chapter 4.............................................................................................. 64
4.7 References ...................................................................................................................... 66
5 Heat-up, devolatilization and combustion of large solid fuel particles.......................... 69
5.1 Introduction to solid fuel combustion ............................................................................ 70
5.1.1 Solid fuel conversion pathways.........................................................................................................72
5.1.2 Heating of solid fuel particles............................................................................................................73
5.1.3 Devolatilization of large fuel particles ..............................................................................................74
5.1.4 Char oxidation...................................................................................................................................75
5.2 Literature study on devolatilization and char combustion kinetics of large particles.... 76
5.2.1 Tire-derived fuels (TDF) ...................................................................................................................76
5.2.2 Wood.................................................................................................................................................77
5.3 Experimental .................................................................................................................. 78
5.3.1 Pre-experimental considerations........................................................................................................78
5.3.2 Experimental set-up...........................................................................................................................79
10
ix
5.3.3 Fuel samples......................................................................................................................................81
5.3.4 Experimental method ........................................................................................................................81
5.3.5 Repeatability and uncertainties..........................................................................................................82
5.4 Results and discussion.................................................................................................... 83
5.4.1 General observations .........................................................................................................................83
5.4.2 Effect of fuel sample mass ................................................................................................................84
5.4.3 Effect of fuel particle size and shape.................................................................................................85
5.4.4 Effect of raw material particle size....................................................................................................86
5.4.5 Effect of raw material bed fill degree................................................................................................87
5.4.6 Effect of oxygen concentration .........................................................................................................89
5.4.7 Effect of rotational speed...................................................................................................................90
5.4.8 Effect of temperature.........................................................................................................................91
5.4.9 Conclusions for experimental parameter study .................................................................................92
5.5 Model analyses............................................................................................................... 93
5.5.1 Model for devolatilization.................................................................................................................93
5.5.2 Validation of devolatilization model .................................................................................................95
5.5.3 Model for char combustion ...............................................................................................................99
5.5.4 Analysis and comparison of char combustion model ......................................................................103
5.6 Conclusions for chapter 5............................................................................................. 105
5.7 References .................................................................................................................... 106
6 Inorganic chemistry in the kiln system............................................................................ 109
6.1 Introduction.................................................................................................................. 109
6.1.1 S, Cl, Na and K species in the rotary kiln, calciner and lower preheater cyclones..........................109
6.2 Literature study on release and capture of inorganic volatile elements ....................... 112
6.2.1 High-temperature reactions between SO
2
and limestone.................................................................112
6.2.2 Removal of chlorine from the kiln system ......................................................................................115
6.3 Equilibrium calculations on influence of reducing conditions..................................... 116
6.3.1 Input data.........................................................................................................................................116
6.3.2 Limitations ......................................................................................................................................117
6.3.3 Influence of reducing conditions .....................................................................................................117
6.3.4 Relative stability of sulfates towards reducing conditions ..............................................................120
6.3.5 Concluding remarks about thermodynamic equilibrium calculations .............................................120
6.4 Experimental ................................................................................................................ 121
6.4.1 Release of SO
2
from calcined raw material .....................................................................................121
6.4.2 Sulfur release from raw materials during combustion of solid fuels ...............................................129
6.5 Conclusions for chapter 6............................................................................................. 139
6.6 References .................................................................................................................... 140
11
x
7 Analysis of industrial experience and combustion model for industrial rotary kilns 143
7.1 Industrial experience with combustion of large fuel particles ..................................... 143
7.1.1 Wet process kilns.............................................................................................................................143
7.1.2 SP kilns............................................................................................................................................144
7.1.3 Calciner kilns...................................................................................................................................145
7.1.4 Conclusions for industrial experience .............................................................................................149
7.2 Modeling combustion of solid fuels in industrial rotary kilns ..................................... 150
7.2.1 Selected full scale model results......................................................................................................152
7.2.2 Conclusions for modeling solid fuel combustion in industrial rotary kilns .....................................155
7.3 Suggestions for alternative fuel utilization................................................................... 156
7.4 Conclusions for chapter 7............................................................................................. 157
References............................................................................................................................... 158
8 Final conclusions ............................................................................................................... 159
8.1 Concluding trends ........................................................................................................ 159
8.2 Suggestions for further work........................................................................................ 160
List of symbols........................................................................................................................... 161
Appendixes................................................................................................................................. 163
Appendix A Sensitivity analysis on estimated fuel conversion........................................... 164
Appendix B Data used in the combustion models............................................................... 167
Appendix C Calculation of values used in the char model .................................................. 169
Appendix D Species included in the equilibrium calculations............................................ 171
Appendix E Gas and bed temperatures through the rotary kiln........................................... 175
Appendix F Stoichiometric calculations on flue gas composition....................................... 177
Appendix G Data used in the industrial model.................................................................... 179
Paper A - Mixing large and small particles in a pilot-scale rotary kiln............................... 181
Paper B High-temperature release of SO
2
from calcined cement raw materials ............ 181
Paper C Sulfur release from cement raw materials during solid fuel combustion.......... 181
Paper D Devolatilization and combustion of tire rubber and pine wood in a pilot scale
rotary drum............................................................................................................................... 181
12
Chapter 1 - Introduction
1
1 Introduction
1.1 Background
Cement production is an energy-intensive process with an energy usage of approximately 3 MJ per
kg cement clinker produced (VDZ, 2009). With an annual global cement production of 3 billion
tonnes cement the cement industry is responsible for approximately 2% of the worlds primary
energy consumption
1
(Cembureau, 2009 and BP, 2010). This corresponds to approximately 10
times the total energy consumption of Denmark (the Danish Energy Agency, 2010).
Traditionally, cement production has mainly depended on the fossil fuels, such as coal, oil and
natural gas. Due to competition in the cement market, increasing fossil fuel prices and
environmental concerns, cement producers have increased the utilization of alternative fuels as a
substitute for fossil fuels, in order to achieve the most economic fuel mix. In this context,
alternative fuels cover all non-fossil fuels and waste from other industries. Secondary, waste or
replacement fuels are also used as synonyms for alternative fuels.
The energy share covered by alternative fuels today is different from country to country, but is most
prominent in Europe and USA, see also chapter 3 section 3.2.1. The reason is mainly that
infrastructure for collecting the fuels is available and the existence of a clear legislation promotes
utilization of alternative fuels in cement production (VDZ, 2009).
Besides these driving forces, the following major advantages are also achieved by utilizing
alternative fuels in cement production:
1. Alternative fuels may be fully or partly CO
2
-neutral.
2. Production costs are typically reduced due to fuel cost savings.
3. Waste from other industries may be recycled as valuable fuel.
4. Ash from the fuels is directly incorporated into the cement clinker.
5. Flame temperatures up to around 2000C and high residence times provide good conditions
for reduction of harmful organic compounds.
However, the following should be considered when alternative fuels are utilized:
1. How is the cement clinker quality affected?
2. How is the process stability and production capacity affected?
3. Is new technology required to combust specific alternative fuels?
4. What is the combustion kinetics for different alternative fuel types?
5. Are emissions of harmful species or other environmental aspects influenced? This could e.g.
be the case by introducing a chlorine by-pass in gas exit from the rotary kiln, which is often
installed in connection with alternative fuel utilization.
6. How are the alternative fuels transported, stored, handled and delivered?
7. Are the alternative fuels available in sufficiently large quantities to justify investments in
required technology and equipment?
1
The primary energy consumption covers energy from coal, oil, natural gas, nuclear and hydro electric power. Other
energy sources such as wind power, solar energy, biomass or geothermal power are not included.
13
Chapter 1 - Introduction
2
The main technical subject of this PhD project is to investigate how solid alternative fuels may be
efficiently utilized in cement rotary kilns without negative impact on clinker quality or process
stability. More specifically, the project is devoted to focus on combustion in the rotary kiln material
inlet end (often called the kiln back-end). The strategy is to first provide a detailed analysis of
current limitations and challenges associated with solid fuel combustion in cement rotary kilns.
Then it is investigated how the inorganic chemistry in rotary kilns may be affected by utilizing
alternative fuels in the material inlet end. Combustion of different alternative fuels in a pilot-scale
rotary drum is conducted to obtain experimental data for validation of a combustion model. Based
on the model, suggestions for modification of the process and technology will be described in order
to optimize solid fuel combustion. Attention should also be given to the work of Larsen (2007)
about alternative fuels in cement production. This work provides an overview of the most relevant
alternative fuels, with particular focus on tire-derived fuels (TDF).
Detailed descriptions of stack emissions of harmful species will not be treated in this thesis. For the
emissions of NO
x
and SO
2
from cement production, attention should be given to the works of
Jensen (1999), Hansen (2003) and Rasmussen (2011). Considerations in terms of economy,
transport, storage, handling and delivery of alternative fuels are also beyond the scope of this thesis.
1.2 Structure
Chapter 2 of this thesis provides introductory information about cement chemistry and production.
It contains process data for a typical, modern cement plant as well as a description of the main
rotary kiln types.
Chapter 3 reviews the current status regarding utilization of alternative fuels in cement production:
Types, amounts and technology. In addition, the chapter will introduce the reader to the main
challenges associated with combustion of alternative fuels in the kiln system.
Chapter 4 provides a basic understanding of the mixing process of particles with different sizes and
densities in rotary kilns. This is partly through a literature study and partly through mixing
experiments in a steel/plexi glass rotary drum.
Chapter 5 describes heat-up, devolatilization and combustion of large solid fuel particles in rotary
kilns. Combustion kinetics for tire-derived fuels and wood is reviewed. Combustion experiments in
a rotary drum under conditions similar to those in the material inlet end of a rotary kiln are
described. The results are examined using mathematical models for devolatilization and char
combustion.
Chapter 6 focuses on inorganic chemistry in the rotary kiln material inlet end. The emphasis is on
how the release of the inorganic volatile elements S, Cl, K and Na is affected by changes in gas
atmosphere towards reducing conditions. Experiments with sulfur release from raw materials during
solid fuel combustion are described.
Chapter 7 provides an analysis of industrial experience with alternative fuels. The analysis is
divided into experience with wet and dry long kilns, suspension preheater (SP) kilns and modern
calciner kilns. The chapter also describes a full-scale model to predict devolatilization and char
combustion times for TDF and other alternative fuels in rotary kilns. Finally, the chapter provides
14
Chapter 1 - Introduction
3
an analysis of different alternative fuels regarding their potential use in the material inlet of rotary
kilns.
Chapter 8 contains the overall conclusions regarding alternative fuel combustion in the rotary kiln
material inlet end. Suggestions for further work are provided.
1.3 Objective
The main objective with this thesis is to provide scientifically based knowledge regarding
combustion of solid alternative fuels in cement production. This is achieved via a combination of
theoretical and experimental investigations coupled with mathematical modeling.
The knowledge may be used to improve design basis, redesign existing technology or to develop
new technology in order to achieve fuel flexible rotary kilns for optimal alternative fuel
combustion. The knowledge may also be used to confirm (or refute) assumptions about fuel
combustion issues in the material inlet end of rotary kilns a location that is very complex to study
experimentally due to the extreme process conditions.
15
Chapter 1 - Introduction
4
1.4 References
BP Statistical Review of World Energy, 2010, http://www.bp.com.
Cembureau, Activity Report 2009, http://www.cembureau.be/
Energistyrelsen, Energistatistik (the Danish Energy Agency, Energy Statistics) 2010,
http://www.ens.dk. (In Danish)
Hansen, J. P.; SO
2
Emissions from Cement Production. PhD Thesis, Technical University of
Denmark, Department of Chemical Engineering, 2003. ISBN 87-90142-96-9.
Jensen, L. S.; NO
x
from cement production Reduction by primary measures. PhD Thesis,
Technical University of Denmark, Department of Chemical Engineering, 1999.
Larsen, M. B.; Alternative Fuels in Cement Production. PhD Thesis, Technical University of
Denmark, Department of Chemical Engineering, 2007. ISBN 978-87-91435-49-8.
Rasmussen, M. H.; Low SO2 emission preheaters for cement production. PhD Thesis, Department
of Chemical and Biochemical Engineering, Technical University of Denmark, 2011.
VDZ, Activity Report 2007-2009, http://www.vdz.de
16
Chapter 2 Cement Chemistry and Production
5
2 Cement Chemistry and Production
This chapter seeks to give the reader a fundamental understanding of cement chemistry and cement
production, which is necessary to read and understand the succeeding chapters.
The section about cement chemistry and processing describes the most important chemical reactions
that raw materials undergo during the formation of cement clinker in the cement plant.
The sections about cement production plants start with a basic overview of a modern cement
production plant. Second, the development of kiln systems from the long wet and dry rotary kilns to
modern calciner kilns is described. Typical process data for a modern calciner kiln are also
described.
2.1 Cement Chemistry and Processing
2.1.1 History
Burnt limestone (mortar) has been used as a binder material for constructions for more than 2,000
years (Peray, 1986). It was early discovered that burning of limestone with certain impurities
improved the quality. Through the years, much experience has been gathered, leading up to the
modern cement production systems that exist in the early 21st Century. Cements can be defined as
adhesive substances capable of uniting fragments or masses of solid matter to a compact whole
(Hewlett, 1998). Cement sets and hardens by chemical interaction with water. Cements are widely
used to produce adhesion between stones, bricks, etc. in the construction of buildings and
engineering works. Today, the main type of cement that is used is Portland cement and it is
primarily based on a mixture of burned limestone, sand and clay. Several other types of cement with
different characteristics can be tailored for specific purposes, but this thesis will only describe the
Portland cement.
2.1.2 Raw materials
The main raw materials used for the Portland cement is naturally occurring limestone, clay and
sand. Limestone is the source of calcium (CaCO
3
, CaMg(CO
3
)
2
), while clay and sand are the
sources of silicon, aluminium and iron (Al
2
O
3
, SiO
2
, Fe
2
O
3
) (Hewlett, 1998). In recent years, there
has been an increasing tendency to use industrial wastes as alternative raw materials. Examples are
coal fly ash from power plants, water treatment sludge or slags from alumina, phosphorus and steel
production (Bhatty et al., 2004).
Solid fuels used for cement production traditional or alternative - introduce a certain amount of
ash components left after the combustion. The ash will also be incorporated in the final cement
clinker. The total mixture of raw materials and ash should match the desired cement clinker
chemistry. Thus, individual evaluations must be performed when either solid fuels or raw materials
are changed.
17
Chapter 2 Cement Chemistry and Production
6
2.1.3 Preparation of raw materials
The raw materials are crushed to particle sizes of approximately 25-100 mm and stored in separate
raw material storages (Andersen, 1981), see Figure 2-1. These stores serve two purposes:
- As intermediate stores to allow cement production to continue even if no new raw materials
are supplied.
- To even out the variations bound to exist in the chemical compositions of the raw materials.
After being reclaimed from their respective stores, the raw materials are mixed in the correct
proportions and transported to the raw mill. The proportions will typically be 75-80 wt.-%
limestone, 20-25 wt.-% clay and possibly smaller amounts of iron ore and/or sand to adjust the
overall composition to meet the requirements for Portland cement (Andersen, 1981).
In the raw mill the mixed raw materials are ground to raw meal (almost as fine as wheat flour), with
a particle size around 5-125 m. The fineness of the particles is important to ensure good heating,
calcining and burning capabilities of the raw materials (Bhatty et al., 2004).
In the raw mill, the raw meal is also dried by hot flue gases from the kiln system. The dried raw
meal ready for production is then transferred to and stored in a silo. The silo serves two purposes:
- To further even out the variations in the chemical composition of the raw meal.
- Act as a buffer store to ensure continuous cement production in cases where no new raw
materials are supplied.
18
Chapter 2 Cement Chemistry and Production

7


Figure 2-1: Typical process flow diagram of a modern cement production plant.

2.1.4 Preheating of raw meal
Raw meal from the silo is led to the top of the preheater tower and heated counter currently by
direct contact with the hot flue gases from the rotary kiln and calciner, see Figure 2-1. Thus the
preheater tower functions as a heat exchanger where the energy from the hot flue gases is
transferred to the raw meal. After having passed the preheater cyclones, the raw meal will have
reached a temperature of approximately 800C (Hewlett, 1998). The preheated raw meal is then
admitted to the calciner where the first major chemical reaction takes place.
The hot flue gases from the rotary kiln may be passed through the calciner (In-Line Calciner, ILC)
or bypassed (Separate-Line Calciner, SLC) before being directed into the preheater cyclones.

2.1.5 Calcination
In the calciner most of the calcium carbonate in the feed will be thermally decomposed to its oxide
and carbon dioxide, see equation (2.1):

CaCO
3
(s) CO
2
(g) + CaO (s) H
25C, 1 atm
= 1782 kJ/kg CaCO
3
(2.1)

19
Chapter 2 Cement Chemistry and Production
8
This very endothermic reaction consumes approximately 60% of the total energy required (Hewlett,
1998). Decomposition of CaCO
3
takes place when the temperature is higher than approximately
700C and continues until all CO
2
is expelled.
2.1.6 Clinker reactions
After having passed the calciner, the calcined raw meal is admitted to the rotary kiln where the
remaining cement clinker reactions take place, see Figure 2-1. Calcined raw meal is also called hot
meal in cement terminology in order to distinguish it from raw meal not yet calcined.
The names, chemical compositions and abbreviations used in cement nomenclature for the four
main constituents of Portland cement clinker are shown in Table 2-1. Belite and alite are the
strength-giving minerals. Alite reacts fast with water during hydration and accounts for the early
strengths while belite reacts slower and gives the cement its late strengths (Hewlett, 1998). Ferrite
and aluminate do not contribute to the strengths of the cement. But they possess important
properties when burning clinker (Hewlett, 1998). These properties of ferrite and aluminate will be
explained later in this section.
Name Chemical composition Nomenclature
Belite 2CaOSiO
2
C
2
S
Alite 3CaOSiO
2
C
3
S
Ferrite 4CaOAl
2
O
3
Fe
2
O
3
C
4
AF
Aluminate 3CaOAl
2
O
3
C
3
A
Table 2-1: Main constituents of Portland cement clinker.
Free lime (CaO), free periclase (MgO), earth alkali sulfates (CaSO
4
, MgSO
4
), Alkali sulfates
(Na
2
SO
4
, K
2
SO
4
) and other minor components may also be found in the clinker (Theisen, 2007).
The requirements for the Portland cement clinker are (Hewlett, 1998):
a) > 67 wt.-% calcium silicates (C
2
S and C
3
S). The remainder must mainly be Fe
2
O
3
, Al
2
O
3
and other oxides.
b) < 5 wt.-% MgO
c) The CaO/SiO
2
ratio by mass shall not be less than 2.0.
Clinker reactions occur at temperatures between 700-1450C, see Figure 2-2. The reactions forming
the final clinker involve intermediate compounds, and the clinker reactions may be affected by
minor compounds. The overall clinker reactions are described, while a detailed overview of the
chemistry, phase relations etc. is provided by Hewlett (1998), Bye (1999) and Bhatty (2004).
20
Chapter 2 Cement Chemistry and Production
9
Figure 2-2 : Phase diagram for cement clinker production (Theisen, 2007).
The formation of belite from CaO and SiO
2
takes place at temperatures above 700C, see equation
(2.2):
2 CaO (s) + SiO
2
(s) 2CaOSiO
2
(s) (2.2)
This reaction is shown in Figure 2-2 where the silicate source quartz and the free CaO is being
consumed while belite is created.
At temperatures above 1,300C, a liquid phase containing alumina (Al
2
O
3
) and iron oxide (Fe
2
O
3
) is
formed. Alumina and iron oxide are not essential to the constitution of the final Portland cement
product but, at this stage in the process, the liquid phase acts as a source of fluxes lowering the
energy requirement thereby making the process economical (Hewlett, 1998). The liquid phase has a
significant influence on the alite formation because it wets all the solid grains, forming an
interpenetrating film. It has two functions (Hewlett, 1998):
- Its surface tension pulls the solid grains together, helping to form nodules known as
clinker.
- It facilitates material transport in the liquid phase.
The liquid phase acts as a media for increased diffusive transport, accelerating the formation of alite
crystals from belite and free lime, see equation (2.3):
CaO (s) + 2CaOSiO
2
(s) 3CaOSiO
2
(s) (2.3)
Aluminate and ferrite will crystallize from the liquid phase when the temperature drops to
approximately 1,230C according to (2.4) and (2.5):
3 CaO (s) + Al
2
O
3
(l) 3CaOAl
2
O
3
(s) (2.4)
4 CaO (s) + Al
2
O
3
(l) + Fe
2
O
3
(l) 4CaOAl
2
O
3
Fe
2
O
3
(s) (2.5)
Liquid phase
Free CaO
CaCO
3
Quartz
Clays
C S
2
C S
3
C A
3
C AF
4
21
Chapter 2 Cement Chemistry and Production
10
The rotary movement of the rotary kiln also helps to create the clinker nodules, by creating a
continuous rolling motion of the clinker material. This rolling motion is described in greater details
in chapter 4. In addition, the inclination of the rotary kiln facilitates a continuous transport of the
calcined raw materials from the rotary kiln material inlet towards the rotary kiln material outlet,
where the clinker nodules are admitted to a clinker cooler.
2.1.7 Clinker Cooling
The Portland cement clinker has reached a temperature of up to 1,450C at the rotary kiln exit, see
Figure 2-1. It is important to recover as much energy as possible and return it to the process. This is
achieved by passing the hot clinker through a grate cooler where typically ambient air will pass the
clinker, thereby cooling them by direct contact in cross current flow. The heated air is used for the
combustion in the rotary kiln and in the calciner.
Experience has shown that it is important with a high rate of clinker cooling from 1,450C to
1,200C if the best strength-giving properties should be achieved. Microscopic investigations have
revealed that a slow cooling from 1,450C to 1,200C often results in alite being partially
transformed back into belite and free lime. This is undesirable since it may lead to setting problems
with the final product (Hewlett, 1998).
2.1.8 Clinker grinding and storage
The clinker leaving the cooler has a variable particle size, typically of diameter 3-25 mm. Thus it is
necessary to grind the clinker to a more uniform particle size. This can either be done directly after
the cooler, or the clinker can be transferred to a silo until it is needed, see Figure 2-1. Cement
clinker is, however, not a stable material: It reacts readily with water/water vapour and CO
2
from
the air. This will result in pre-hydration and carbonation. Extended storage periods may therefore
have an effect on the cement properties.
In connection with the clinker grinding, 3-6 wt-% gypsum (CaSO
4
2H
2
O) is added to the clinker.
The gypsum has a marked effect on both the strength and the setting of cement (Theisen, 2007).
Additives such as coal fly ash, sand or raw material may also be added at this point in order to
contribute positively to the strength-giving properties of the cement (Hewlett, 1998). Some of these
additives can replace a significant fraction of the clinker, thereby saving energy needed for
calcination and clinker reactions. Thus the additives have a great potential to reduce energy
consumption and CO
2
-emissions per kg cement produced (WWF, 2008). For this reason, many
cement companies seek to minimize their environmental impact by exploring the use of new
Supplementary Cementitious Materials (SCM), examples being volcanic ashes or kaolinite clays
(Ballan, 2011).
The grinding of clinker and additives can be performed with several different mill systems. When a
sufficient fineness is reached, the final Portland cement product is transferred to a cement silo. It is
stored in this silo until it is being packed and shipped to the end users.
22
Chapter 2 Cement Chemistry an
2.2 The kiln system
The production of cement may be
processing and 3) clinker process
treatment of the raw materials ne
in the preheater, calciner, rotary k
system. This section will aim to d
the calciner or the rotary kiln part
2.2.1 Rotary kiln
The rotary kiln is often referred to
clinker formation reactions take p
of an outer steel shell and an inne
for modern rotary kilns are betwe
3 and rotate 1-5 rpm in order to
as nodulization. Production capac
may be as high as 12,000 tpd (FL
In the material outlet, the rotary k
form a flame to provide energy fo
burner should be short, narrow an
flame to the materials in the bed (
The hot material bed is seen belo
Modern rotary kiln burners are of
using multi-channel burners with
to adjust primary air amounts, inj
(Vaccaro, 2006, FLSmidth, 2011
The recent trend in multiple fuel b
flexibility towards fuel particle si
Figure 2-3: Left: Outer view of a rotary
seen next to the rotary kiln, see also sec
nd Production
11
e divided into three parts: 1) Preparation of ra
sing, storage and shipment. Pyro-processing c
cessary to obtain the cement clinker. Pyro-pro
kiln and cooler. These sections are commonly
describe the modern kiln system since alternat
t of the kiln system.
o as the heart of every cement plant. This is w
place. The rotary kiln is simply a long, cylind
er refractory lining, see Figure 2-3. Typical le
een 40-100 m, and 3-6 m, respectively. Rotary
facilitate mass transport and ensure clinker fo
city is typically 2,000-4,000 tonnes of clinker
LSmidth, 2009).
kiln is equipped with a burner. The burners m
or the clinker reactions to take place. The flam
nd strongly radiant in order to achieve a good
(Vaccaro, 2006). Figure 2-3 illustrates the kil
w the burner.
ften designed to burn a variety of fuels. This i
h separate channels for fuels and primary air w
jection velocities and momentum independen
). Swirl may be used to enhance mixing and s
burners is to use a single common fuel chann
ize and type (FLSmidth, 2009).
y kiln seen from above. Dimensions (LxD) = 40 m x 4
tion 2.2.4. Right: Inner view of a rotary kiln seen from
aw materials, 2) pyro-
covers the thermal
ocessing takes place
y referred to as the kiln
tive fuels are fired in
where the chemical
drical tube consisting
engths and diameters
y kilns are inclined 1-
orming processes such
r per day (tpd), but
main function is to
me of the rotary kiln
d heat transfer from the
ln burner and flame.
is today achieved by
which make it possible
ntly of the fuel flows
stabilize the flame.
nel which allows more
m. The tertiary air duct is
m the burner end.
23
Chapter 2 Cement Chemistry and Production
12
In early cement plants, the kiln system consisted only of a rotary kiln: Raw materials were dried,
preheated, calcined and burned to clinker on their way through the rotary kiln. This process required
very long rotary kilns, often significantly longer than 100 m. The raw materials were either
introduced as dry raw meal or as water/raw material slurry, and this type of plants were therefore
commonly referred to as dry long kilns or wet long kilns, respectively. Due to a low energy
efficiency, this type of cement plants is very expensive to operate, and are rarely constructed today.
However, many wet and dry long kilns still exist, especially in North America (Bhatty, 2004).
The rotary kiln consists of an outer steel shell and an inner refractory lining for thermal insulation,
in order to maintain and resist the high process temperatures. In a rotary kiln, the refractory usually
consists of bricks of special composition and sizes, able to withstand high temperatures. However,
the refractory may also be a cast lining of concrete. The refractory lining is subject to a wide range
of destructive influences through the mechanical dynamics of the rotary kiln, the chemistry of the
cement clinker process and the type of fuels used. The intensity of these stresses varies according to
the operating conditions and kiln sections (Bartha, 2004). The rotary kiln is therefore equipped with
a range of refractory bricks with different properties to ensure appropriate kiln zone lining.
The burning zone refractory lining usually suffers the greatest wear due to the higher temperatures
in the burning zone. However, the burning zone lining is protected by a coating layer which prolong
the lifetime of the refractory lining. The coating is a mass of clinker or dust particles that adheres to
the wall of the kiln, having changed from a liquid or semiliquid to a solidified state (Peray, 1986).
Generally, the burning zone refractory lifetime is 9 to 12 months depending on the specific kiln type
and operating conditions. The refractory lifetime of the colder rotary kiln material inlet zone is
typically 12 to 48 months. Thus, these different rotary kiln zones do not have to be replaced quite as
frequently (Bartha, 2004).
The most used brick types today are chromium-free magnesia-alumina-spinel bricks with MgO
content of 80-95 wt.-% and Al
2
O
3
content of 3-18 wt.-%. Minor compounds are typically Fe
2
O
3
,
Mn
2
O
3
, SiO
2
, CaO and ZrO
2
(Bartha, 2004 and Harder, 2008). These brick types are regarded as
having the longest service life and the best price/performance ratio (Harder, 2008).
The increased use of alternative fuels in the cement industry may lead to higher levels of
recirculating alkali metals and sulfur within the kiln system, see chapter 3, section 3.4.4.
Recirculating alkali metals and sulfur cause large quantities of salts to condense in and on the
refractory lining, predominantly in the temperature window 750C to 1,100C (Harder, 2008). Salt
compounds enter into reactions with refractory bricks that contain alumina and the bricks can be
destroyed by salt crystallization and alkali spalling. Sulfur oxides make the reactions even worse, by
formation of alkali sulfate salts (Klischat et al., 2002).
Various strategies are used by the refractories industry to counteract these wear processes. These
include use of additives that produce low gas permeability and reduce the infiltration tendency of
alkalis. Other solutions are sealing or impregnating the refractory material to form a protective zone
(Harder, 2008). One of the most successful and widespread solutions is addition of 3-6 wt.-%
silicon carbide, SiC, which leads to an appreciable resistance to alkali attack (Klischat et al., 2002,
Bartha, 2004 and Harder, 2008). The addition of SiC leads in situ to formation of liquid phases,
which seal the refractory surface and protect against alkali infiltration (Bartha, 2004).
Increased circulation of inorganic volatiles such as sulfur and chlorine in the kiln system, due to
increased alternative fuel utilization also entails a higher risk of kiln shell corrosion (Jns and
24
Chapter 2 Cement Chemistry and Production
13
stergaard, 2001). Efforts have been made to identify suitable refractory steels for cement rotary
kilns, with characteristics that are a compromise between good creep resistance, high corrosion
resistance in the presence of chlorine and sulfur, and strong resistance to abrasion when hot
(Corcuera, 2004).
2.2.2 Design of modern rotary kilns for ILC systems
This section will outline important design criteria for modern rotary kilns. Since these design
criteria may differ between different kiln systems, focus will be on the In-Line Calciner (ILC) type
which is described in greater detail in section 2.2.4. Cement plants of the ILC type are among the
most commonly built cement plants in recent years, and this trend seems to continue.
The rotary kiln dimensions depend on the required production capacity and of the thermal load. The
production capacity is related to the mass and energy balances through the rotary kiln because the
production capacity is a function of the hot flue gas volume passing through the rotary kiln per time
unit. The thermal load can be expressed by the two terms a) volumetric load and b) burning zone
load.
The volumetric load is a measure of the production in tpd clinker per m
3
kiln volume inside lining,
and has the unit tpd/m
3
. The volumetric load decreases when the kiln dimensions increases. Typical
volumetric loads for ILC type rotary kilns are between 3.6 and 5.3 tpd/m
3
. However, the volumetric
load can be higher if the burnability of the raw materials is good.
The burning zone load is defined as energy quantity per time per area of burning zone, and has the
units W/m
2
. If the kiln dimensions are increased, the burning zone load will decrease because the
energy quantity per time unit will be distributed over a larger area. Typical burning zone loads for
ILC rotary kilns are 2.9 to 6.2 MW/m
2
. The burning zone load must be kept between these specific
limits in order to ensure a sufficient refractory lining lifetime.
Modern rotary kilns may be supported by either two or three supports. The 3-support rotary kilns
typically have a length-diameter (L/D) ratio of 16 to 18 whereas 2-support rotary kilns typically
have L/D ratio of 10-14 (FLSmidth, 2005). A 2-support rotary kiln is shown in Figure 2-4. The
material inlet end is shown to the right and the grate cooler is shown to the left.
Each support has two rollers which support the rotary kiln and facilitates its rotary movement. Only
one pair of supporting rollers is connected to a kiln drive which rotates the entire rotary kiln. The
other pair of rollers acts simply as supports for the rotary kiln. To ensure optimal contact between
the rollers and the rotary kiln, self-adjusting supports have been developed which can equalize the
load between the rollers, also under varying operating conditions (FLSmidth, 2005).
Both ends of the rotary kiln are equipped with lamella seals to prevent hot gas and solid materials
from escaping from the kiln system. Lamella seals are a relatively new invention which has
successfully replaced traditional pneumatic and spring seals. Lamella seals are practically
maintenance-free (FLSmidth, 2005).
25
Chapter 2 Cement Chemistry an
Figure 2-4: 2-support rotary kiln. Each
2-support rotary kilns have severa
kiln support reduces the total init
The shorter rotary kiln dimension
It is well-known that kiln shells m
situations that involves temporary
deformation may cause loss of co
on the other rollers. In compariso
four supporting rollers.
2-support rotary kilns usually hav
production rate and volumetric lo
load results in improved burning
One drawback with the shorter L/
temperatures of the flue gases lea
calcined raw meal fed to the rotar
to avoid deposit build-ups in the k
control temperatures by feeding r
duct (FLSmidth, 2005).
Regarding alternative fuel combu
of the rotary kiln in order to ensu
2.2.3 Preheater and calciner s
In modern cement plants, raw me
preheater, see Figure 2-5. The raw
preheater. Heat exchange with th
cyclone (the top cyclone). The ga
and finally to the stack. In this wa
the gas between the 2
nd
and 3
rd
st
separation. The solid residence tim
seconds (Strau et al., 1987).
Most of the calcination process ta
designed to decompose CaCO
3
to
emissions to a minimum.
nd Production
14
support has two rollers, one of them with kiln drives (
al advantages compared to 3-support rotary k
ial installation cost, and fewer components ne
ns also lead to a lower heat loss from the kiln
may suffer deformation, e.g. due to short kiln
y cooling-heating cycles. For a 3-support rota
ontact with one or two of the supporting roller
on, the 2-support rotary kiln will always main
ve a larger diameter than 3-support rotary kiln
oad, resulting in a lower burning zone load. T
zone refractory lifetime (FLSmidth, 2005).
/D ratio of a 2-support rotary kiln is that it ma
aving the rotary kiln, depending on calcining
ry kiln, raw meal burnability and other proces
kiln riser duct caused by high temperatures, i
raw meal from the second-lowest preheater cy
ustion in the rotary kiln, it may be necessary t
ure sufficient time for fuel burnout and ash sin
systems
eal is preheated to calcination temperature in a
w material feed is fed to the system between t
e gas takes place followed by a gas/solid sepa
as that leaves the 1
st
stage is transferred to the
ay, as much heat as possible is recycled. The
age where they experience a similar heat exch
me through a 5-stage cyclone preheater is in t
akes place in a separately fired calciner. The c
o CaO, but also to convert the fuel sufficiently
(FLSmidth, 2005).
kilns: Eliminating one
eed to be maintained.
surface.
shutdowns or other
ary kiln such
rs and increased loads
ntain contact with all
ns of the same
he lower burning zone
ay lead to higher
degree of the partially
ss conditions. In order
it may be necessary to
yclone to the kiln riser
to increase the length
ntering.
a multi -stage cyclone
the 1
st
and 2
nd
cyclone
aration in the 1
st
stage
e raw mill, dust filter
solids are conveyed to
hange and gas/solid
the range of 50
calciner is not only
y and to reduce
26
Chapter 2 Cement Chemistry and Production
15
There are different kinds of calciner systems, but the two main types are:
1) In-line calciner (ILC)
2) Separate line calciner - Downdraft (SLC-D)
The ILC system may also be connected to separate equipment for combustion of solid alternative
fuels, such as the HOTDISC. This will be described in greater detail in chapter 3, section 3.3. In the
following, a representative description of an ILC system will be given.
2.2.4 In-Line Calciner (ILC) system
The ILC system shown in Figure 2-5 is characterized in that the flue gas from the rotary kiln is led
directly through the calciner. The degree of calcination in the ILC system is very high, often in the
range of 90-95%. The production capacity is in the range of 1,500-12,000 tpd and the ratio of firing
in the calciner is typically 55-65%.
Figure 2-5: In-Line Calciner (ILC) system. Right: Preheater cyclones and calciner shown in 3D view (FLSmidth,
2011).
A part of the combustion air in the calciner is drawn directly from the cooler through a separate
tertiary air duct between the cooler and the calciner. Thus, the gas volume through the rotary kiln is
reduced and the size of the rotary kiln can be smaller. In addition, the ILC system provides good
conditions for reduction of NO
x
formed in the rotary kiln due to reducing conditions in the region
between the calciner fuel injection and the tertiary air admission (Jensen, 1999). The reducing zone
is illustrated in Figure 2-6, which also shows the tertiary air duct above the reducing zone. Above
the fuel inlet and tertiary air duct is an oxidizing zone where the fuel is combusted, leading to an
increase in temperature. This temperature increase is controlled by addition of raw meal to the
lower calciner part, which uses the heat for the calcination reaction. Above the restriction is the
Air
Clinker
~100C
Tertiary air
850-1100C
~290C
~480C
~650C
~800C
~890C
Calciner fuel
Rotary
kiln fuel
Raw meal inlet
Rotary kiln
Calciner
Preheater
27
Chapter 2 Cement Chemistry and Production

16

calcining zone where most of the endothermic calcination reaction takes place, facilitated by an
efficient heat transfer from gas to solids. Gas and solid particles leaves the calciner through a loop
duct, which provides good conditions for CO oxidation by improving the gas mixing.



Figure 2-6: The In-Line-Calciner. Notice the NO
x
-reducing zone in the lower calciner part (Snderborg, 2007).

The ILC system is also well suited for alternative fuels in solids forms, because they burn in
suspension imaging single particle behavior. The suspension type burning ensures good external
mass and heat transfer (Larsen, 2007).
The temperature in the ILC system reflects the equilibrium temperature of calcination and is
typically 850-900C. In the lower part, the oxygen level is low because fuel is admitted without its
combustion air. In the region of the tertiary air duct, preheated air increases the oxygen content to
typically 11-12% O
2
. The oxygen level decreases from this point and up to the lowermost cyclone
due to the combustion process and due to a dilution with formed CO
2
. In modern FLSmidth
calciners the gas velocity through the calciner is normally kept at 8 m/s and the gas retention time is
about 3-4 seconds before the loop duct starts (Snderborg, 2007).


Fuel
Tertiary air
Loop duct for
gas mixing
Reducing zone
Raw meal
Raw meal
Oxidizing
zone
Rotary kiln
28
Chapter 2 Cement Chemistry and Production
17
2.3 Overview of process data
This section provides an overview of important process data for a modern calciner kiln. The given
process data are representative of an ILC system.
Temperatures and retention times for both gas and solids are shown in Table 2-2. From this table it
is observed that the solid material is heated from approximately 100C in the preheater to 1,450C
in the rotary kiln burning zone, and then finally cooled to approximately 100C in the clinker
cooler. The gas being transported in the opposite direction is observed to enter the system at
ambient temperature in the clinker cooler (or as primary air through the kiln burner). The gas then
reaches temperatures of up to around 2,000C in the rotary kiln burning zone whereafter the gas
temperature gradually decreases through the rotary kiln, calciner and preheater before it leaves the
kiln system at temperatures around 350C. The gas is then used to preheat the cement raw materials
in the raw mill, and will finally leave through the stack at temperatures around 100C. It should be
noted that gas and material retention times are typical values, but will depend on the specific
cement plant layout.
Equipment Preheater Calciner Rotary kiln Cooler
Gas temperature (C) 350-880 880-1100 1100-2000 1100-ambient
Gas retention time - 5-10 s - 3-4 - 5-10 s - 1 s
Material temperature (C) 100-780 780-900 900-1450 1400-100
Material retention time - 50 s - 5 s 15-30 min - 30 min
Table 2-2: Typical temperatures and residence times in a modern calciner kiln (Cembureau, 1999).
Oxygen concentrations in a modern calciner kiln
Clinker cooler and rotary kiln
The combustion air enters the kiln system through either the clinker cooler or kiln burner as ambient
air with oxygen concentration around 21 vol.-%. In the rotary kiln the oxygen concentration
decreases as the combustion proceeds to approximately 2-7 vol.-% in the rotary kiln gas outlet.
However, a fraction of the combustion air is bypassed via the tertiary air duct and will be
transported directly from the clinker cooler to the calciner as approximately 1,000C hot,
atmospheric air.
Calciner
In the lower part of the calciner, the oxygen level is low because fuel is admitted without its
combustion air, see Figure 2-6. In the region of the tertiary air duct, preheated air increases the
oxygen content to typically 11-12 vol.-% O
2
. The oxygen level decreases from this point and up to
the lowermost cyclone due to the combustion process and due to a dilution with formed CO
2
.
Preheater
In the preheater section, the oxygen concentration is only slightly affected by oxidation reactions of
minor components in the raw meal, e.g.:
1. Formation of SO
2
from pyrite (FeS
2
) which is the main reason for SO
2
emissions from
cement plants (Hansen, 2003).
2. Recarbonation of CaO forming CaCO
3
.
3. Combustion of organic bound carbon.
29
Chapter 2 Cement Chemistry and Production
18
In addition, small amounts of atmospheric air (false air) are often introduced into the preheater
section due to the slightly negative pressure.
Stack
Oxygen levels in the stack flue gas are typically around 5-10 vol.-%.
2.4 Conclusions for chapter 2
Production of Portland cement requires that the raw materials limestone, clay and sand are heated
up to approximately 1,450C in order to first calcine limestone and, secondly, facilitate the clinker
reactions between lime and silicates. The main strength giving components in Portland cement are
belite (C
2
S) and alite (C
3
S).
The production of cement may be divided into three parts: 1) Preparation of raw materials, 2) pyro-
processing and 3) Clinker processing, storage and shipment. Pyro-processing covers the thermal
treatment of the raw materials necessary to obtain the cement clinker. Pyro-processing in a modern
cement plant takes place in the preheater, calciner, rotary kiln and cooler. These sections are
commonly referred to as the kiln system.
Since the calcination is a very endothermic process, 50-60% of the total energy is usually fired into
the calciner, whereas the remaining 40-50% is fired through the kiln burner. However, some cement
plants do also fire a part of the energy into the rotary kiln material inlet end or through separate
combustion devices coupled to the kiln system. This will be described in greater detail in chapters 3
and 7.
Temperatures in the kiln system range from ambient temperatures up to 2,000C, and oxygen levels
range from 2 vol.-% to 21 vol.-% dependent on position in the kiln system.
Solid and gas retention times vary in the different sections of the kiln system. In the rotary kiln, gas
retention times may be in the order of 5-10 seconds, while solid retention times may be as long as
15-30 minutes.
30
Chapter 2 Cement Chemistry and Production
19
2.5 References
Andersen, P. S.; The Cement Factory; FLSmidth & Co. A/S, Valby, Denmark, 1981.
Ballan, J.; Refining pozzolan the greener alternative. Highlights, FLSmidth, 32-33, May 2011.
Bartha, P.; The cement rotary kiln and its refractory lining. Interceram, Refractories Manual, 14-17,
2004.
Bhatty, J. I., Miller, F. McG. and Kosmatka, S. H. (Editors); Innovations in Portland Cement
Manufacturing, Portland Cement Association (PCA), Illinois, USA, 2004. ISBN: 0-89312-234-3.
Bye, G. C.; Portland Cement; 2
nd
edition, Thomas Telford, London, 1999. ISBN 0-7277-2766-4.
Cembureau; Best Available Techniques for the Cement Industry, Belgium, 1999. Available from:
http://cembureau.be/.
Corcuera, A.; Refractory steels: AFs and corrosion problems. World Cement, 99-102, November,
2004.
FLSmidth; Burning bright, World Cement, 101-102, December 2009.
FLSmidth; Preheater calciner systems, Brochure, FLSmidth, Valby, Denmark, 2011. Available
from: http://flsmidth.com.
FLSmidth; ROTAX-2 Kiln components, Brochure, FLSmidth, Valby, Denmark, 2005. Available
from: http://flsmidth.com.
FLSmidth; Fast payback burner upgrades. International Cement Review, 121-122, May 2011.
Hansen, J. P.; SO
2
Emissions from Cement Production. PhD Thesis, Technical University of
Denmark, Department of Chemical Engineering, 2003. ISBN 87-90142-96-9.
Harder, J.; Trends in refractory materials for the cement industry. ZKG International, Vol. 61, No.
5, 36-45, 2008.
Hewlett, P. C. (Editor); Chemistry of Cement and Concrete; 4
th
edition, John Wiley & Sons Inc.,
New York, 1998. ISBN 0-340-56589-6.
Jensen, L. S.; NO
x
from cement production Reduction by primary measures. PhD Thesis,
Technical University of Denmark, Department of Chemical Engineering, 1999.
Jns, E. S. and stergaard, M. J.; Kiln Shell Corrosion, IEEE-IAS/PCA Cement Industry Technical
Conference, Vancouver, British Columbia, Canada, 343-359, 2001.
31
Chapter 2 Cement Chemistry and Production
20
Klischat, H.-J., Liever, H. and Wirsing, H.; Alkali-resistant linings for the security and preheating
zones in rotary cement and lime kilns subject to chemical attack. ZKG International, No. 6, 66-75,
2002.
Larsen, M. B.; Alternative Fuels in Cement Production; PhD Thesis, Department of Chemical
Engineering, Technical University of Denmark, 2007. ISBN 978-87-91435-49-8.
Peray, K. E.; The Rotary Cement Kiln; 2
nd
edition, Edward Arnold Ltd., Victoria, Australia, 1986.
ISBN 0-7131-3609X.
Strau, F., Steinbi, E. and Wolter, A.; Measurement of retention times in cement burning systems
with the aid of radionuclides. ZKG-International, 9, 441-446, 1987.
Snderborg, H. R.; The International Cement Production Seminar 2007, Volume IA, Lecture 05-01;
FLSmidth A/S, Valby, Denmark, 2007.
Theisen, K.; The International Cement Production Seminar 2007, Volume II, Lecture 06-02/03;
FLSmidth A/S, Valby, Denmark, 2007.
Vaccaro, M.; Burning alternative fuels in rotary cement kilns, IEEE Conference Paper, 127-136,
2006.
WWF, A blueprint for a climate friendly cement industry. WWF International, Gland, Switzerland,
2008.
32
Chapter 3 Alternative fuels in cement production
21
3 Alternative fuels in cement production
This chapter provides an overall insight into the use of alternative fuels in cement production.
Section 3.1 describes the general requirements for alternative fuels to be used, and section 3.2
reviews current trends in the cement industry. Section 3.3 describes technology for combustion of
alternative fuels, advantages as well as disadvantages. Finally, section 3.4 provides an overview of
challenges related to the substitution of fossil fuels with alternative fuels.
3.1 Requirements for alternative fuels in cement production
The global cement production in 2009 was estimated to be 3 billion tonnes (Cembureau, 2009). The
fuel energy consumption is about 3 MJ/kg cement (Schneider, 2009), corresponding to a total global
energy consumption of approximately 9 10
12
MJ/year. It is estimated that the cement industry is
responsible for approximately 2% of the worlds primary energy consumption, corresponding to
approximately 10 times the total energy consumption of Denmark (BP, 2010 and ENS, 2010). It is
also estimated that cement production accounts for 5% of the global CO
2
-emissions, arising partly
from fuel combustion, thermal calcination of limestone, electricity use and transport (CSI, 2005;
PBL, 2007).
Fuel consumption accounts for a significant fraction of the total cement clinker production costs,
which makes it attractive for cement producers to consider cheaper substitute fuels. In addition,
increasing concerns about CO
2
-emissions put pressure on the cement industry to reduce emissions,
e.g. by substituting fossil fuels with CO
2
-neutral fuels. However, the cement industry has certain
requirements with respect to alternative fuels, the most important ones being:
1. Overall clinker production costs.
2. Availability (must be available in sufficiently large amounts).
3. Legislation (a clear legislation and permission from local authorities is often mandatory
before specific alternative fuels can be used).
4. Technical possibility for transport, handling and combustion of the specific alternative fuel.
In addition to these general requirements for alternative fuels, there are some fuel specific
requirements as well:
1. The content of water (as low as possible).
2. Content of minor components must be within acceptable limits.
3. Emissions, process stability and product quality must not be affected negatively.
4. Alternative fuel size and shape.
5. Energy density.
Re 1: An increase in water content compared to traditional fuels can increase the flue gas volume.
This may decrease the production capacity of the cement kiln if ID-fan capacity is insufficient
(Krennbauer, 2006). The change in production capacity depends mainly on the water content of the
fuel. Pre-drying of the fuel can help to reduce this problem, or the cement plant can be equipped
with an ID-fan with a sufficiently large capacity to compensate for an increase in exhaust gas
33
Chapter 3 Alternative fuels in cement production
22
volume. Too high moisture contents of the alternative fuel may also slow the drying and
combustion process and potentially move the combustion zone in e.g. the calciner.
Re 2: Most chemical elements from the fuel ash are incorporated into the cement clinker. However,
volatile elements such as chlorine, sodium, potassium, sulfur, mercury and thallium will partly
evaporate in the hotter regions of the kiln system. These volatile elements may condense from the
gas phase again in e.g. the preheater, being captured in the bag filter, or being removed through a
bypass or escape with the stack gases. In order to ensure the cement quality and process stability, it
is therefore necessary to control the levels of all chemical inputs via the fuel. This will be further
explained in section 3.4.
Re 3: Substitution to a specific alternative fuel must not have any negative effect on the emission
level, process stability or product quality. Such considerations will be described further in section
3.4.
Re 4: The combustion behavior of solid fuel particles depends on the particle size and shape. If
there are large variations in fuel particle size and shape, it becomes challenging to control the
cement production. For this reason, there are often requirements to the maximum allowable fuel
particle size, and sometimes also shape, to be used in a specific kiln system. Please note here that
the virgin fuel shape may change considerably upon combustion, see also section 5.1.1.
Re 5: It is generally desirable to have as high an energy density as possible, in order to reduce
transportation costs. Some alternative fuels have relatively low energy densities. Consequently,
larger fractions must be fed to the kiln system to give the same energy as fuels with a higher energy
density. These low energy density fuels may still be attractive to use, if the price is low enough to
balance the additional costs associated with transport, handling, etc.
Many of the alternative fuels that are available for the cement industry in large amounts consist of
mixtures of waste with fluctuations in composition. Thus, alternative fuels are often relatively in-
homogenous fuels. Since modern cement kiln systems are always designed to operate with a
continuous clinker production, fluctuations in fuel quality may occur, necessitating a stringent
process control.
34
Chapter 3 Alternative fuels in cement production
23
3.2 Types and quantities of alternative fuels
A study of the various types of alternative fuels has been provided by Larsen (2007) and Schneider
(2009). It is evident from these studies that the amounts of alternative fuels are increasing for each
year. The substitution rate is particularly high in Europe, Japan and in North America. This is
mainly due to the fact that these regions have developed the necessary infrastructure and handling
facilities as well as having established clear legislations with respect to landfills, waste incineration,
emission levels, etc. (CSI, 2005; Sutou and Yukio, 2000).
3.2.1 Energy share by alternative fuels
The thermal energy share covered by alternative fuels in cement production varies from region to
region. One of the leading regions is Germany which had a thermal share covered by alternative
fuels of 54% in 2008 (Schneider, 2009). This is an average thermal share for 50 German cement
plants. A few individual cement plants are reported to have alternative fuels energy shares in the
range 90-100% (Scheuer, 2003). Chinese cement producers, which were responsible for 54% of the
global cement production in 2009 utilizes practically no alternative fuels as yet, but interest in
alternative fuel utilization is on the rise (Cembureau, 2009 and Chang-Ming, 2007).
Table 3-1 shows the thermal energy share from alternative fuels reported by a number of cement
producers, responsible for approximately 20% of the global cement production. The thermal energy
share is observed to be in the range of 0.7% to 22% with typical values around 12%. All of these
cement producers have a common goal to increase their use of alternative fuels in the future. ICR
Research has also recently estimated the global average alternative fuel share to be around 12%,
which corresponds well with the figures reported here (ICR Research, 2011).
Cement producer %Thermal energy
share covered by
alternative fuels
Global share of cement
production in %
(million tonnes)
Year Reference
Lafarge 12 4.5 (136) 2010 Lafarge, 2010
Holcim 12 4.4 (132) 2009 Holcim, 2009
CEMEX 20 3.2 (96)
1
2010 CEMEX, 2010
Heidelberg Cement 22 2.6 (78) 2010 Heidelberg, 2010
Taiheiyo Cement 12 1.4 (43) 2010 Tayheiyo, 2010
Grasim 0.8 1.1 (33) 2008 Grasim, 2008
SCG 14 1.0 (32) 2010 SCG, 2010
ACC 0.7
2
0.7 (21) 2010 ACC, 2010
Siam City Cement 5 0.5 (15) 2009 Siam City, 2010
CRH 14
2
0.4 (13) 2009 CRH, 2009
Cementir 6 0.3 (10) 2009 Cementir, 2009
Table 3-1: Thermal energy share covered by alternative fuels by cement producer (average values for all the cement
producers cement plants).
1
This figure is the annual production capacity and not the actual annual production.
2
This
figure includes used oil as alternative fuel.
35
Chapter 3 Alternative fuels in cement production
24
3.2.2 Types and quantities of alternative fuels
Many types of alternative fuels are applied in the cement industry. Figure 3-1 shows the alternative
fuel types and amounts in the German cement industry in 2010. This figure gives an indication of
the diversity with respect to physical and chemical properties between the different fuels. The
majority of the alternative fuels are in solid form, while liquids are less common.
Figure 3-1: Alternative fuels used in German cement production in 2010 (VDZ, 2010).
Combustion in cement kiln systems provides society with a useful waste management option that
can be an economically viable and environmentally sound alternative to land disposal, treatment or
incineration. In recent years, the mad cow disease has created vast amounts of meat and bone
meal which have successfully been destroyed in a number of cement kilns. In some countries, e.g.
Norway, national policy makes cement kilns the preferred solution for hazardous waste
management (CSI, 2005).
The type of alternative fuel deployed depends on local conditions and may differ substantially from
country to country. In Germany for example, there is a tendency to increased use of processed
commercial and industrial waste due to the EC directive 1999/31/EC which bans the landfill of
unprocessed waste (Schneider, 2009). This type of waste is often referred to as Refuse Derived
Fuel, RDF. Other important alternative fuels in German cement industry are meat and bone meal,
tire-derived fuels, waste oil and processed municipal waste.
Another example is in Switzerland, where the Swiss pharmaceutical industry produces large
amounts of organic waste solvents every year. These waste solvents are popular alternative fuels in
the cement industry due to their usually high energy content and because they are easy to handle. In
2002, 30,600 tonnes of waste solvents were used by the Swiss cement industry, representing a share
of approximately 14% of the total energy consumption (Seyler et al., 2005).
3.2.3 Fuel analysis
In this section, fuel analysis for some commonly used alternative fuels are presented, see Table 3-2.
The values are approximate, and may deviate depending on source, age, pre-treatment, etc.
However, the values give an impression of heating values and elemental compositions in the
36
Chapter 3 Alternative fuels in cement production
25
different fuels. In addition, the fraction of volatile matter and fixed carbon are important for the
combustion characteristics.
The volatile content of the most commonly deployed alternative fuels is observed to be in the range
54-100 wt. %. This is significantly higher than for coals or petcoke used in cement production,
being 6-38 wt. %. Consequently, the char content is generally quite low, 0-30 wt. % for alternative
fuels whereas the char content in coal and petcoke is in the range 50-90 wt. % (Jensen, 1999).
The ash content of some alternative fuels, e.g. refuse derived fuels (RDF), meat and bone meal
(MBM) or tire-derived fuel (TDF), may be as high as one-third of the total mass. The high ash
content may be problematic if the ash contains a large fraction of elements which may affect the
cement quality adversely. This could for example be phosphorus in MBM or zinc in TDF. However,
the ash may also contain elements that can contribute positively to the process: Calcium, silicon,
alumina or iron from the ash will reduce the need of raw materials, making the cement production
more economic.
Biomass, here presented in the form of pine wood, wheat straw and rice husks, are seen to have
quite similar elemental composition and lower heating value of around 15-20 MJ/kg. However, both
lower heating value and content of ash, water as well as elements such as sulfur, potassium, sodium
and chlorine may differ significantly depending on the specific biomass type.
RDF is an inhomogeneous mixture of processed commercial and industrial waste. Lower heating
value and elemental composition may vary significantly depending on location, but lower heating
values are often in the range of 14 MJ/kg (Tokheim, 2005). The content of particularly chlorine is
often relatively high. The high chlorine content needs to be considered if RDF is to be used as
alternative fuel, see section 3.4.
TDF is a widely used alternative fuel in cement production and is one of those alternative fuels that
are most similar to coals: The lower heating value is around 31-33 MJ/kg and coals are around 30-
35 MJ/kg (Jensen, 1999). The similarity exists also with respect to content of volatile matter, char
and sulfur.
MBM has lower heating values in the range of 16-20 MJ/kg. The sulfur and chlorine contents of
MBM may be up to 1 wt.-% and 0.6 wt.-%, respectively.
Plastics, here presented as poly ethylene, PE, have the highest lower heating values of 40-44 MJ/kg.
They consist of almost 100 wt. % volatile matter, making them similar to oils. For this reason,
plastics are generally considered as excellent alternative fuels. An exception, however, is poly-vinyl
chloride, PVC, which contains approximately 57 wt. % chlorine. Due to the high chlorine content,
the heating value is around 19 MJ/kg, less than half the value of other common plastics. The high
chlorine content is also undesirable for process reasons, which will be described in section 3.4.
Sewage sludge consists mainly of volatiles and ash, with fixed carbon contents typically less than 6
wt.-%. The lower heating value is around 14-19 MJ/kg which is quite low. Sewage sludge may also
contain relatively high amounts of sulfur and chlorine, up to 0.8-1 wt.-%.
37
C
h
a
p
t
e
r

3


A
l
t
e
r
n
a
t
i
v
e

f
u
e
l
s

i
n

c
e
m
e
n
t

p
r
o
d
u
c
t
i
o
n


2
6

T
a
b
l
e

3
-
2
:

F
u
e
l

a
n
a
l
y
s
e
s

a
n
d

h
e
a
t
i
n
g

v
a
l
u
e
s

o
f

a
l
t
e
r
n
a
t
i
v
e

f
u
e
l
s

c
o
m
m
o
n
l
y

u
s
e
d

i
n

t
h
e

c
e
m
e
n
t

p
r
o
d
u
c
t
i
o
n
.

U
n
i
t
s

a
r
e

d
r
y

w
t
.
%

a
n
d

M
J
/
k
g

d
r
y

b
a
s
i
s
.

N
o
t
e
s

a
r
e
:

a
)

V
o
l
a
t
i
l
e

m
a
t
t
e
r
,

b
)

F
i
x
e
d

c
a
r
b
o
n

(
c
h
a
r
)
,

c
)

P
i
n
e

w
o
o
d
,

d
)

R
e
f
u
s
e

D
e
r
i
v
e
d

f
u
e
l
,

e
)

T
i
r
e
-
d
e
r
i
v
e
d

F
u
e
l
,

f
)

M
e
a
t

a
n
d

B
o
n
e

M
e
a
l
,

g
)

P
o
l
y
-
E
t
h
y
l
e
n
e
,

h
)

P
o
l
y
-
V
i
n
y
l
-
C
h
l
o
r
i
d
e
,

i
)

b
y

d
i
f
f
e
r
e
n
c
e
,

j
)

l
o
w
e
r

h
e
a
t
i
n
g

v
a
l
u
e

c
a
l
c
u
l
a
t
e
d

f
r
o
m

t
h
e

h
i
g
h
e
r

h
e
a
t
i
n
g

v
a
l
u
e
,

k
)

S
e
w
a
g
e

S
l
u
d
g
e
.

P
r
o
x
i
m
a
t
e

a
n
a
l
y
s
i
s

U
l
t
i
m
a
t
e

a
n
a
l
y
s
i
s


L
o
w
e
r

h
e
a
t
i
n
g




V
M
a

F
C
b

A
s
h

C

H

N

S

C
l

O
i

N
a

K

v
a
l
u
e

R
e
f
e
r
e
n
c
e

W
o
o
d
c

7
9
-
8
3

1
5
-
1
7

0
.
3
-
0
.
5

4
9
-
5
1

5
.
8
-
6
.
0

0
.
0
6
-
0
.
0
7

0
.
0
1
-
0
.
0
3

-

3
6
-
4
4

-

-

1
9
.
7
-
1
9
.
8
j

H
a
s
s
a
n

e
t

a
l
.
,

2
0
0
9

G
a
u
r

a
n
d

R
e
e
d
,

1
9
9
8

W
h
e
a
t

s
t
r
a
w

7
4
.
5

1
2
.
9

6
.
8

4
5
.
9

5
.
9
6

1
.
1

0
.
2

0
.
6
3

3
9
.
4

0
.
0
2

1
.
8
1

1
8
.
1
j

S
t
e
n
s
e
n
g
,

2
0
0
1

R
i
c
e

h
u
s
k
s

6
1
-
6
9

1
6
-
1
7

1
5
-
2
0

3
8
-
4
2

3
.
0
-
4
.
8

0
.
4
-
0
.
6

0
.
1
-
0
.
2

-

3
6
-
5
4

0
.
1
1

2
.
1
8

1
5
.
4
j

C
h
e
n

e
t

a
l
.
,

2
0
1
1

V

r
g
e
h
y
i

e
t

a
l
.
,

2
0
1
1

K
a
r
m
a
k
a

e
t

a
l
.
,

2
0
1
1

R
D
F
d

6
0
-
8
4

2
-
1
6

8
-
3
2

4
2
-
7
2

5
-
1
1

0
.
8
-
2
.
4

0
.
1
-
0
.
6

0
.
1
-
3
.
9

3
-
3
6

0
.
1
-
0
.
2

0
-
0
.
1

1
4

T
o
k
h
e
i
m
,

2
0
0
5

a
n
d

K
o
b
y
a
s
h
i

e
t

a
l
.
,

2
0
0
5

T
D
F
e

5
4
-
6
6

2
3
-
3
0

7
-
2
3

6
4
-
8
1

5
.
6
-
7
.
2

0
.
3
-
0
.
5

1
.
4
-
1
.
6

0
.
1
5

2
.
0
-
5
.
6

~

0

~

0

3
1
.
0
-
3
2
.
8

C
h
e
n

e
t

a
l
.
,

2
0
0
1

L
a
r
s
e
n
,

2
0
0
7

C
h
i
n
y
a
m
a

e
t

a
l
.
,

2
0
0
7

M
B
M
f

6
5
-
7
2

7
.
2
-
9
.
7

1
7
.
5
-
2
8
.
3

4
2
.
1
-
4
5
.
9

5
.
8
-
6
.
4

7
.
5
-
1
1
.
2

0
.
3
8
-
0
.
9
7

0
.
3
5
-
0
.
5
8

1
5
.
3
-
3
8
.
4

0
.
3
1

0
.
0
7

1
6
.
2
-
1
9
.
9

A
h
o
,

M
.

e
t

a
l
.
,

2
0
0
5

A
y
l
l
o
n
,

M
.

e
t

a
l
.
,

2
0
0
6

K

n
t
e
e
,

U
.

e
t

a
l
,

2
0
0
4

P
E
g

1
0
0

~

0

~

0

8
6

1
4

~

0

~

0

~

0

~

0

~

0

~

0

4
0
.
2
-
4
4
.
2
j

P
a
n
a
g
i
o
t
o
u
,

T
.
,

1
9
9
4

W
a
n
g
,

Z
.

e
t

a
l
,

2
0
0
4

P
V
C
h

9
1

9

1

3
8

5

~

0

~

0

5
7

~

0

~

0

~

0

1
9
.
2
j

P
a
n
a
g
i
o
t
o
u
,

T
.
,

1
9
9
4

W
a
n
g
,

Z
.

e
t

a
l
,

2
0
0
4

S
S
k

4
8
-
6
4

0
-
6

1
7
-
5
0

2
8
-
3
9

3
.
8
-
4
.
1

3
.
8
-
4
.
3

~

1

0
.
1
-
0
.
8

2
9
-
3
6

0
.
2
4

0
.
6
3

1
4
.
0
-
1
9
.
1

M
e
a
s
u
r
e
d

38
Chapter 3 Alternative fuels in cement production
27
3.3 Technologies for alternative fuels combustion
The traditional firing points in a cement plant are the calciner and kiln burner, see chapter 2. With
the introduction of alternative fuels these two units have been subject to modifications in order to
optimize the combustion. However, is it still required to use relatively small fuel particles to ensure
a) the desired flame profile in the kiln burner and b) sufficient fuel burnout in the calciner.
Since many alternative fuels are costly to downsize, it is often desirable to utilize coarse alternative
fuels directly, thereby minimizing the shredding expense. This has led to the development of
various forms of extensional equipment for combustion of lumpy alternative fuels. Selected types of
extensional equipment will be described in this section.
3.3.1 HOTDISC
The HOTDISC from FLSmidth is a combustion chamber designed for combustion of large solid
fuels, such as whole tires, automobile dashboards and upholstery, wood waste and RDF (Keefe,
2003, FLSmidth, 2008, 2009 and 2011). It is a large, rotating hearth furnace, integrated with the
preheater and calciner. Figure 3-2 shows the HOTDISC and its typical position in a calciner kiln.
The combustion air is hot tertiary air from the cooler. Raw meal from the lower cyclone stages of
the preheater is added to control temperatures inside the HOTDISC. It is possible to add cold raw
meal in order to control the temperature in the HOTDISC.
Figure 3-2: Left: A cross-sectional view of the HOTDISC. Right: The position of the HOTDISC in a calciner kiln
(FLSmidth, 2011).
Calciner
Cold raw
meal bin
Triple flap sluice
Raw meal inlet
Tertiary air
HOTDISC
39
Chapter 3 Alternative fuels in cement production
28
In the HOTDISC, it is possible to adjust the disc rotational speed to ensure sufficient residence time
for fuel burnout. The fuel is transported approximately 270 degrees on the disc before it reaches the
scraper, where ash, partially calcined raw meal and possible unburned fuel will be discharged into
the lower calciner or riser duct.
Five HOTDISC units are in operation as of early 2011 and three new HOTDISC units are under
installation. In 2008, the HOTDISC technology was awarded the Global Fuels Award in the
category: Most innovative technology for alternative fuel use.
3.3.2 The Pre-Combustion Chamber (PCC)
According to Schmidthals (2003) the Pre-Combustion Chamber (PCC) is designed for large fuel
particles of wood and TDF. The PCC is shown in Figure 3-3 next to the calciner. In the PCC,
thermal conversion of large fuels take place under strongly sub-stoichiometric conditions ( - 0.2)
producing a lean gas that is used as gaseous fuel in the calciner. The remaining solids are conveyed
by a discharge system into the kiln inlet via a rectangular discharge channel. Entrainable fractions
of char and ash are lifted into the calciner, while larger solid fractions, e.g. steel wires from tires,
drop into the kiln inlet (Schmidthals, 2003).
Figure 3-3: Left: PCC shown next to the calciner. Right: Bottom-to-top view inside the PCC (ThyssenKrupp, 2002).
Rotary kiln
Pre-Combustion
Chamber
Alternative fuel inlet
Calciner
Cooled tertiary air
(combustion air)
Product gas outlet
Product gas outlet
Combustion air inlet
Alternative fuels inlet
40
Chapter 3 Alternative fuels in cement production
29
3.3.3 Gasifiers
Gasifiers for gasification of alternative fuels are in use at some cement plants (Wirthwein et al.,
2002 and Saito et al., 1987). The gasifiers produce a gaseous fuel that is used in the calciner while
the solid residues are used as raw materials for cement production.
A circulating fluidized bed gasifier is installed at a German cement plant and use waste from the
paper industry as well as packing and plastic waste as alternative fuels. It has a capacity of up to 20-
25 t/h of alternative fuels, producing about 50,000 Nm
3
/h of gas with a heating value around 3,000-
5,000 kJ/Nm
3
. This is sufficient to cover approximately 40% of the total energy consumption
(Wirthwein et al., 2002).
So-called refuse energy gasifiers (REGs) utilize whole tires. The energy recovery from the gasifiers
is reported to be as high as 95%. To the authors knowledge, two REGs are in operation today at
Japanese cement plants. The tire capacities are 300-400 kg/h and 500-800 kg/h, respectively. This
corresponds to overall fuel substitution rates of 5-7% and 11-18%, respectively (Saito et al., 1987).
3.3.4 Kiln material inlet firing and mid-kiln firing
Several cement plants have experience with combustion of lumpy alternative fuels in either the
material inlet end of rotary kilns or through the kiln shell (mid-kiln firing). Firing directly into the
rotary kiln is attractive given that it requires no additional combustion devices. But as will be
explained in further details in subsequent sections, as well as in chapters 6 and 7, there are
challenges associated with combustion processes in the rotary kiln, where solid fuels and cement
raw materials are in physical contact. Still, some cement plants do utilize up to 20% of the total
thermal energy by either kiln material inlet or mid-kiln firing. Full-scale experience with lumpy
alternative fuel combustion in rotary kilns is the subject of chapter 7.
There are a number of technologies for alternative fuel combustion in the rotary kiln material inlet
end and/or mid-kiln. Some solutions are illustrated in Figure 3-4: To the left is shown a feed shelf
system which allows tires to be fired into the kiln material inlet. Above the feed shelf is seen a
fork which holds the tire in the calciner, while the tire-devolatilizes/combusts. The tire is then
released, whereby solid residues in the form of steel wires or unburned char are dropped into the
rotary kiln. On the right of Figure 3-4 is shown an example of mid-kiln firing: Every time the rotary
kiln has rotated one revolution, a tire is dropped through a flap damper and into the rotary kiln.
Mid-kiln firing is predominantly used in long wet and long dry rotary kilns.
Figure 3-4: Solutions for back-end firing (left) and mid-kiln firing (right) (Cadence Recycling, 2009).
41


Chapter 3 Alternative fuels in cement production
30
3.4 Effect of alternative fuels on process stability and product quality
The utilization of alternative fuels in cement production may alter the input of minor elements
introduced to the kiln system, since some alternative fuels have a different chemical composition
than traditional fossil fuels. It is important to be aware of the effect of these minor elements as they
may affect the stability of the kiln system or clinker quality. This section will describe the influence
of minor elements on the kiln system and clinker quality.
3.4.1 Cement chemistry
A number of elements may be introduced with the alternative fuels, and could affect the clinker
production. Such components are typically:
1. Alkali metals (potassium and sodium).
2. Sulfur
3. Chlorine
4. Zinc
5. Phosphorous
6. Iron
7. Unburned carbon in the material charge.
8. Quartz (Silicon)
Re 1-2. Alkali metals (sodium and potassium) and sulfur can affect the cement chemistry as well as
the kiln system stability. High alkali levels in the final cement can react with moisture and produce
a gel which expands and gives rise to cracking in concretes (Hewlett, 1998). The alkali content of
the cement can be controlled through careful selection of the raw materials or by extracting high-
alkali dust from the kiln system.
When sufficient sulfur is present in the clinker, the alkalis are normally present as sulfates such as
K
2
SO
4
, Na
2
SO
4
, Na
2
SO
4
3K
2
SO
4
(aphthitalite), (CaOSiO
2
)
2
CaSO
4
(sulfate-spurrite) and/or
2CaSO
4
K
2
SO
4
(langbeinite). In the absence of sufficient sulfate in the clinker, Na
2
O can enter the
C
3
A and increase the reactivity, potentially leading to setting problems. K
2
O will normally enter the
C
2
S where it enhances the reactivity to water and may inhibit its conversion to C
3
S. A reducing
atmosphere in the kiln is reported to facilitate the transfer of alkalis from the sulfate forms to the
C
3
A and C
2
S (Hewlett, 1998).
When the alkalis are present as sulfates in the burning zone of the kiln system, the viscosity of
liquid phase is decreased, thus promoting the formation of C
3
S (Hewlett, 1998).
The term alkali equivalent, Na
2
O
e
, is normally used to describe the amount of alkali present in the
clinker product:
Na
2
O
e
=
2
2
2 2
% %
Na O
K O
MW
wt K O wt Na O
MW
+ = 0.66 wtK
2
O + wt%Na
2
O (3.1)
Higher alkali levels in cement (> 0.8wt% Na
2
O
e
) have the effect of increasing the early strength of
the cement at the expense of the late strength (Hewlett, 1998).
Excess amounts of sulfur tend to form CaSO
4
. The amount of sulfur available for forming CaSO
4
is
termed excess SO
3
and may be calculated by the empirical formula:
42
Chapter 3 Alternative fuels in cement production
31
Excess SO
3
= Total SO
3
( )
( )
( )
( )
3 3
2 2
2 2
1
( 2
W W
W W
M SO M SO
K O Na O
M K O M Na O

= Total SO
3
0.85 K
2
O 0.65 Na
2
O (3.2)
Where total SO
3
, K
2
O and Na
2
O are the clinker concentrations in g/kg clinker, and are determined
by a chemical analysis. Analyses of different clinker compositions have revealed that nearly the
entire amount of potassium and half the amount of sodium forms sulfate (Hansen, 2003).
It is highly desirable for as much as possible of the alkalis and sulfur to be discharged from the kiln
system with the clinker. Otherwise, some of the alkalis and sulfur are volatilized in the high
temperature zones of the rotary kiln and condense as sticky, low-melting point solids in the colder
parts. This can cause blockage problems known as coating and ring formation. Condensation of
sticky solids also serves to attract dust and bind it, which gives further rise to blockage problems
(Hewlett, 1998). This is explained in further details in section 3.4.4.
Re 3. Chlorides readily volatilize in the burning zone and condense in the cooler parts to combine
with alkalis and sulfates to form low melting point mixtures. Their effect upon the operation of kiln
systems is so severe that it has often been necessary to either set an upper limit for the chlorine
content of the raw materials/fuel, or to introduce a chlorine bypass in the kiln/preheater system. A
bypass can be very effective to remove chlorides from the kiln system. Operation with a chlorine
bypass is described in greater detail in chapter 6. The chlorine content in Portland cements is
typically restricted to 0.1 wt% in order to avoid serious problems of reinforcement corrosion in
concrete (Hewlett, 1998). Some cement manufacturers also produce low chlorine cements.
Re 4. Small amounts of zinc (0.01-0.2 wt.-%) have been reported to increase the reactivity of C
3
A
and may give rise to time-setting problems (Hewlett, 1998). This may reduce the use of tire-derived
fuels (TDF) in cement production, since TDF contains a variable amount of zinc (1.00-1.25 wt.-%).
If the total energy share covered by TDF approaches 30%, the quantities of zinc in the clinker
reaches the acceptable limit (Pipilikaki, P. et al., 2005).
Re 5. Phosphorus is normally present in the clinker as P
2
O
5
in a range of 0.03 to 0.22 wt%.
Amounts of P
2
O
5
higher than 0.5-1.0 wt% stabilize C
2
S to an extent where conversion to C
3
S is
inhibited. The addition of small amounts of fluorine-containing compounds prevents this effect
(Hewlett, 1998). P
2
O
5
is particularly present in meat and bone meal (MBM), but may also be
present in household waste.
Re 6. Tire-derived fuels (TDF) contain a relatively large amount of steel (up to 25 wt% for truck
tires) (Norcini et al., 1998). During the combustion process, TDF will thus contribute as an iron
source, thereby lowering the demand for iron sources in the raw material mixture.
Re 7. Incomplete burnout of the fuel may cause reducing conditions in the material charge in the
kiln. These reducing conditions may increase the emissions of SO
2
as well as damage the refractory
lining in the kiln. This will be explained in further details in section 3.4.4. Reducing conditions may
also produce so-called brown clinker which differs in color from traditional Portland cement.
43
Chapter 3 Alternative fuels in cement production
32
Re 8. Sewage sludge contains silicates (quartz sand) with coarser grains than normally seen in
cement raw meal. Coarser silicate particles may react with already formed alite to produce belite,
thereby resulting in lower cement strength. Use of large quantities of sewage sludge may therefore
be harmful to cement quality.
Trace elements such as As, Cd, Cr, Cu, Ni and V from the alternative fuels will be incorporated into
the clinker. Due to their small quantities, these trace elements normally do not affect the clinker
quality in any measurable degree. The hazardous risk from these elements can also be considered to
be of minor importance because the trace elements are firmly incorporated into the crystal phases
during cement hydration (Wanzura, 2003).
3.4.2 Emissions
Generally, assuming a correct equipment design, the use of alternative fuels will not lead to
increased emissions, compared to fossil fuels (Karstensen, 2008, Lee et al., 2007, Lemarchand,
2000, Prisciandaro et al., 2003, Realff et al., 2005 and Wurst, 2003). However, emissions from by-
pass should be considered because a bypass is most often needed when using alternative fuels. The
application of alternative fuels may interact with the process conditions in the bypass extraction
area with attendant risk of bypass emissions. This section will briefly review how stack emissions
are affected when alternative fuels are used in cement production.
3.4.2.1 Legislation
Cement plants located in the European Union which use alternative fuels are subject to Directive
2010/75/EU about industrial emissions (Directive 2010/75/EU). Important emission limits are
summarized in Table 3-3.
Pollutant Emission limit Unit
Dust 30 mg/Nm
3
@ 10 vol-% dry
SO
2
50
1
mg/Nm
3
@ 10 vol-% dry
NO
x
500 mg/Nm
3
@ 10 vol-% dry
HCl 10 mg/Nm
3
@ 10 vol-% dry
HF 1 mg/Nm
3
@ 10 vol-% dry
Hg and Tl 0.05 mg/Nm
3
@ 10 vol-% dry
Dioxins and furans 0.1
1
TE
a
/Nm
3
@ 10 vol-% dry
TOC 10 mg/Nm
3
@ 10 vol-% dry
CO Set by local authorities
Table 3-3: Emission limits for co-incineration of alternative fuels in cement kilns. a: TE = Toxic Equivalents (Directive
2010/75/EU).
1
Exemptions may be authorized by the competent authority in cases where SO
2
and TOC arises from
others sources than the fuel, e.g. from the raw materials.
Older cement plants in the EU may be allowed to have higher NO
x
emissions than 500 mg/Nm
3
, but
only if they fulfil the requirements mentioned in a BAT (Best Available Technique) reference
document for the lime, cement and magnesium oxide manufacturing industries written by the
European Parliament (European Commission, 2010).
Each pollutant, except dust, presented in Table 3-3 will be briefly described in the following
sections.
44
Chapter 3 Alternative fuels in cement production
33
Emissions may differentiate significantly between individual cement plants. For this reason,
emission data from European cement plants have since 2001 been registered in the European
Pollutant Release and Transfer Register (E-PRTR). Emission data will be evaluated for each cement
plant with a production capacity above 500 tpd, and regularly published on the internet (European
Commission, 2011).
3.4.2.2 SO
2
SO
2
emissions from modern cement plants originate mainly from oxidation of pyrite, Fe
2
S, or
organically bound sulfur in the upper preheater stages. SO
2
originating from the fuel is to a great
extent absorbed by limestone in the calciner or lower preheater stages (Hansen, 2003). An exception
is, however, cement plants operating with a bypass: The flue gas purged through this bypass is
normally a) released to the atmosphere through a separate stack or the main stack, thereby giving
rise to SO
2
emissions, or b) recirculated to the preheater, calciner or clinker cooler (Heidelberg
Cement, 2010). SO
2
emissions from cement plants have attracted much attention in recent years,
and reference is made for the works of Hansen (2003) and Rasmussen (2011).
3.4.2.3 NOx
NO
x
emissions from cement plants may arise from either thermal NO
x
or from nitrogen in fuels. A
detailed description of NO
x
formation and reduction is given by Jensen (1999). In addition, Larsen
(2007) reviewed the effect of co-firing alternative fuels on NO
x
-emissions. He found that NO
x
emissions generally decreased when alternative fuels were used.
The emission limits of NO
x
of 500 mg/Nm
3
have made it necessary for new cement plants to use
NO
x
reduction measures such as staged combustion, described in chapter 2, section 2.2.4, or the
SNCR technique (VDZ, 2005).
3.4.2.4 HCl and HF
Emissions of acidic gases such as HCl and HF are generally very low due to the alkaline conditions
in the kiln system: Chlorine and fluorine are predominantly bound as salts rather than HCl and HF
(VDZ, 2005). However, cement plants operating with a chlorine bypass have been reported to emit
HCl and HF (Heidelberg Cement, 2010).
3.4.2.5 Hg and Tl
Mercury and thallium are highly volatile heavy metals that are able to escape through the stack as
opposed to other heavy metals, which are primarily incorporated into the clinker (Hewlett, 1998).
Stack emissions of particularly mercury from cement plants as well as power plants and waste
incineration plants, have attracted much attention in recent years. For detailed information about
mercury release and capture, reference is made to the works of Paone (2008) and Zheng (2011).
3.4.2.6 Dioxins and furans
Dioxins and furans cover a wide range of poly-chlorinated organic compounds, of which many are
toxic. Since the toxicity of individual dioxins and furans fluctuates, the toxicity is defined in the
unit Toxic Equivalents per Nm
3
, TEQ/Nm
3
. 1 TEQ/Nm
3
is defined as the toxicity for the dioxin
45
Chapter 3 Alternative fuels in cement production
34
2,3,7,8-tetrachlorodibenzodioxin (TCDD), and the toxicity of all other dioxins and furans are
evaluated relative to TCDD (Directive 2010/75/EU).
A review article about formation, release and control of dioxins and furans from cement kilns are
provided by Karstensen (2008). Dioxin and furan emissions from cement kilns are generally very
low, because oxygen concentrations, temperatures and residence time are sufficient to ensure full
conversion of organic species to their ultimate combustion products CO
2
and H
2
O.
Formation of dioxins and furans are typically caused by improper mixing of fuel and combustion air
in the calciner, where residence time and temperature are lower than in the rotary kiln (Karstensen,
2008). For cement plants operating with a bypass, it is important to keep the bypass filter
temperature below 250C since higher temperatures may generate dioxins and furans (Wirthwein,
2007).
3.4.2.7 CO and TOC Total Organic Carbon
Alternative fuels have different combustion characteristics than fossil fuels, which may influence
the CO formation mechanism. CO oxidation rates drop drastically with temperature in the
temperature window 650C to 950C and almost no CO oxidation takes place below 650C
(Larsen, 2011). It is therefore important that the calciner is designed to oxidize CO before it escapes
to the preheater.
Overall, CO and TOC emissions from the kiln system may be caused by four main factors (Larsen,
2011):
1) Incomplete mixing of fuel and oxidizer in the calciner system.
2) Fluctuations in fuel feed rate.
3) Incomplete fuel burnout in the calciner system.
4) Organic carbon in the raw materials.
Regarding the first three bullet points, CO emissions may be minimized by a proper equipment
design and operation to match the specific fuels. The fourth bullet point is not connected with
alternative fuels but depends on the organic carbon content of the raw materials.
3.4.3 Local reducing conditions
It is widely accepted in the cement industry that cement clinker should be burned in an oxidizi ng
atmosphere. Sub-stoichiometric oxygen concentrations in the kiln system cause deterioration of
clinker quality and affect kiln process stability negatively (Hewlett, 1998). This section will
describe how alternative fuels may cause local reducing conditions in the kiln system, and why
local reducing conditions have a negative impact on clinker quality and process stability.
3.4.3.1 Local reducing conditions caused by incomplete fuel combustion
Incomplete combustion of solid fuels may lead to unburned fuel dropping into the material charge
in the rotary kiln. This is illustrated in Figure 3-5 where unburned fuel can reach the material charge
in four different ways:
46
Chapter 3 Alternative fuels in cement production

35

1) Coarse solid fuels fired into the calciner may drop through the riser duct and into the rotary
kiln.
2) Coarse solid fuels fired through the kiln burner may not be completely combusted in the
flame, and will consequently fall into the material charge.
3) Solid fuels fired into the calciner may be light enough to be carried upwards through the
calciner and into the lower cyclone stage. But if the combustion is incomplete, unburned
carbon may be transported with calcined raw meal from the lower cyclone stage and into the
rotary kiln material inlet end.
4) Direct firing of, typically large, solid fuels into the rotary kiln material inlet end.


Figure 3-5: Solid fuels dropping into the material charge in the rotary kiln.

Partially burned or unburned alternative fuels in the material inlet end of the rotary kiln can
consume oxygen and thereby cause reducing conditions. Reducing conditions in the material charge
do not necessarily mean reducing conditions in the gas phase.
Reducing conditions in the material charge may also occur if the kiln burner flame shape or
direction is such that the flame impinges on the material charge and consumes all oxygen near the
material charge.

3.4.3.2 Effects of local reducing conditions on the clinker quality
It is widely accepted that Portland cement should be burned under oxidizing combustion conditions
in order to ensure a good product quality (Klauss, 2000). The main effect of clinker burning under
reducing conditions is mainly due to the raw material component Fe
2
O
3
(Fe(III)) being reduced to
Fuel and
primary air
Calciner fuel
Cooling air
Calciner
To preheater
Rotary kiln
4) Coarse
alternative
fuel particles
Cement clinker
Tertiary air
Cyclone
1)
2)
3)
47
Chapter 3 Alternative fuels in cement production
36
FeO (Fe(II)) or, in the case of strong reducing conditions, elemental Fe (Fe(0)). Reduction of Fe(III)
under reducing conditions take place at temperatures above 450C. Fe(II) affects the clinker quality
for the following reasons (Klauss, 2000):
- Fe(II) catalyzes alite decomposition (3CaOSiO
2
2CaOSiO
2
+ CaO), the main strength-
giving component in cement.
- Less ferrite (4CaOAl
2
O
3
Fe
2
O
3
) is formed. Instead, more aluminate (3CaOAl
2
O
3
) is
formed, which may lead to altered setting during the hydration of cement mortar.
In addition to these effects on cement quality, the color of the cement may also change. This is
commonly known as brown clinker.
3.4.3.3 Effects of local reducing conditions on process stability
Local reducing conditions in the material charge in either the rotary kiln material inlet end or in the
kiln burning zone have an effect on the process stability of the entire kiln system. The process
stability is mainly affected by increased release of sulfur from the raw meal to the gas phase, which
may be summarized by the overall reaction:
CaSO
4
(s) + CO (g) CaO (s) CO
2
(g) + SO
2
(g) (3.3)
Similar reactions with alkali sulfates are possible, releasing both sulfur and alkali metals to the gas
phase. Chapter 6 describes the reductive decomposition of sulfates in greater detail.
3.4.4 Alkali, sulfur and chlorine: Circulation in the kiln system
This section provides overall insight into the volatilization of Cl, S, K and Na from the raw
materials and fuel during the combustion process and clinker reactions in the kiln system.
A cycle of volatiles that evaporate in the hot part of the rotary kiln and condense in the cold part of
the rotary kiln, calciner and preheater, trap the volatiles in the kiln system, causing accumulation of
volatiles within the circulation loop. This phenomenon is well-known in the cement industry.
Circulation of chlorine in the kiln system is mainly related to high temperatures in the rotary kiln
and is hardly affected by the kiln atmosphere (Locher and Klein, 2009). Circulation of alkali metals
and sulfur is related to both high temperatures and the kiln atmosphere: Alkali metals and sulfur
present in the partially calcined raw meal as sulfates are sensitive to reducing conditions which
favors release to the gas phase. Particularly sulfur bound as CaSO
4
(commonly referred to as excess
sulfur) is sensitive to reducing conditions (Nievoll et al., 2009). The reactions leading to sulfur and
alkali release is described in further details in chapter 6.
Circulation of large amounts of the volatile elements Cl, Na, K and S are problematic because these
elements may a) condense and form eutectic salt mixtures of alkali sulfates and alkali chlorides,
and/or b) react with clinker minerals and form complex compounds. Both options can lead to sticky
deposit build-ups in the rotary kiln, riser duct and lower cyclone stages. The melting point of the
eutectic salt mixtures depends on the composition, but can be well below 900C (Palmer, 1990).
Deposit build-ups inside the rotary kiln are often called kiln rings. Kiln rings are more severe build-
ups than the coating layer that normally adhere to the internal walls in the rotary kiln. While the
48
Chapter 3 Alternative fuels in cement production
37
coating layer consisting of cement raw materials is desirable because it protect the refractory bricks
from erosion and reduces heat loss to the surrounding atmosphere. Kiln rings induce operational
instability and restrict gas and material flow to a point where production must be stopped (Palmer,
1990). Kiln rings in the material inlet end can be formed for several reasons besides circulation of
volatiles. These other reasons may be frequent changes in chemical composition and fineness of
raw meal, excessive dust generation within the rotary kiln, frequent variations in temperatures in the
freeboard gas through the kiln length, etc. Peray (1986) gives a detailed description of the reasons
and mechanisms behind ring formation.
The chemical composition of the kiln ring depends on the location along the rotary kiln length due
to variations in temperature and kiln atmosphere (Palmer, 1990). Investigations of kiln rings from
several industrial rotary kilns have confirmed that sulphospurrite (2C
2
SCaSO
4
) is a main ring-
building mineral phase, with typical concentrations in the range of 25-30% (Nievoll et al., 2009).
Sylla (1974) found that besides sulphospurrite was also spurrite (2C
2
SCaCO
3
) and calcium
sulphoaluminate (C
3
A
3
CaSO
4
) frequently found in kiln rings. Sylla reported that spurrite was
primarily formed in CO
2
-rich atmospheres at temperatures between 700-900C. Spurrite formation
was observed to be favored by the presence of chlorides and alkalies. Sulphospurrite and calcium
sulphoaluminate was stable at the temperature intervals 900-1,150C and 950-1,300C,
respectively. The stability range of sulphospurrite and calcium sulphoaluminate was observed to
increase if the SO
2
concentration in the kiln gas increased.
Potgieter and Wirth (1996) reported that the mechanism of kiln ring formation was related to the
presence of either a low melting CaO-Al
2
O
3
-SiO
2
liquid phase or a (Na, K)
2
O containing liquid
phase. The liquid phase acts as a binding agent for building up the ring on the refractory lining. In
particular the high concentration of alkali liquid phase increases the stickiness and thus enhances
adhesion of dust and raw meal which becomes incorporated into the kiln ring.
Katsioti et al. (1995) studied the formation of coatings in the lowest cyclone stage and in the rotary
kiln inlet. The coatings in the lowest cyclone stage were found to contain high levels of spurrite and
chlorides, mainly KCl. The coatings in the kiln inlet contained only high levels of spurrite, MgO
and CaO. The coatings in both positions were explained to be caused by circulation of alkali metals
and chlorine in the kiln system. A maximum limit of 500 ppm chlorine in the raw meal entering the
rotary kiln was recommended in order to reduce coating formation in the lowest cyclone stage.
Sutou et al. (1999) reported that coating formation in the lower cyclone stages and kiln riser duct
was caused by the interaction of CaO and SiO
2
with chlorine and sulfur circulating in the kiln
system. Particularly the low-melting point minerals sulphospurrite (2C
2
SCaSO
4
) and
chloroellestadite (3C
2
S3CaSO
4
CaCl
2
) were reported to be problematic.
Circulation of volatiles in the kiln system may also lead to corrosion of rotary kiln shells. Jns and
stergaard (2001) studied kiln shell corrosion in 12 suspension calciner kilns. They found that
corrosion was particularly severe when the kiln gas contained high levels of both sulfur and
chlorine. In addition, increased sulfur circulation may lead to increased calciner corrosion. This may
be caused by condensation of sulfuric acid on the inner steel shell.
49
Chapter 3 Alternative fuels in cement production
38
3.5 Conclusion
The use of alternative fuels in the cement industry has increased significantly during the last 20
years, and this tendency will continue in the future due to strong drivers such as economic and
environmental concerns. Some of the most important alternative fuels are solid fuels such as tire-
derived fuels, refuse derived fuels, meat and bone meal and waste wood. These fuels have physical
and chemical characteristics different from traditional, fossil fuels, which create a need for modified
or new combustion equipment.
Since many of the solid, alternative fuels are expensive to down-size, it is attractive to be able to
feed these fuels as relatively coarse fuel particles. This may be problematic in the traditional firing
points in the kiln main burner or calciner, which have led to the development of extensional
equipment that is specially designed for large solid fuel particles. The HOTDISC is an example of
such specially designed equipment. Another firing point which may be attractive for firing of coarse
alternative fuels is the material inlet end of rotary kilns. It is relatively straightforward to feed fuels
into this location, and high temperatures and long retention time in the rotary kiln provide good
conditions for fuel burn out. A challenge, however, may be that direct, physical contact between
fuels and raw meal can affect the overall process stability of the kiln system or maybe the product
quality.
Some types of alternative fuels contain relatively high levels of volatile inorganic elements such as
chlorine, sulfur and alkali metals, and may potentially increase the amount of these volatile
elements that circulates in the kiln system. In addition, local reducing conditions in the material
charge caused by insufficient fuel-air mixing can lead to increased release of sulfur and alkali
metals from the raw materials. And finally, the fuel combustion may increase local temperatures in
the rotary kiln which also lead to increased release of sulfur, alkali metals and chlorine from raw
materials.
All in all, solid fuels fired into the material end of rotary kilns have the potential of increasing the
circulation of inorganic volatiles in the kiln system. Increased circulation of inorganic volatiles is
problematic because it accelerates deposit formation in the rotary kiln, riser duct and lower cyclone
stages.
50
Chapter 3 Alternative fuels in cement production
39
3.6 References
ACC; Sustainable Development Report 2010 Web Update. ACC Limited, Mumbai, India, 2010.
Aho, M. and Ferrer, E.; Importance of coal ash composition in protecting the boiler against chlorine
deposition during combustion of chlorine-rich biomass, Fuel, 84, 201-212, 2005.
Aylln, M., Aznar, M., Snchez, J. L., Gea, G. and Arauzo, J.; Influence of temperature and heating
rate on the fixed bed pyrolysis of meat and bone meal, Chem. Eng. Journal, 121, 85-96, 2006.
BP Statistical Review of World Energy, 2010, http://www.bp.com.
Cadence Recycling, 2009. http://www.cadencerecycling.com.
Cembureau, Activity Report 2009, Belgium.
Cementir, Environmental Report 2009, Cementir Holding, Rome, Italy, 2009.
Cemex, CEMEX Annual Report 2010, Mexico, 2010.
Chang-Ming, G. Domestic cement industry production profit function between waste and gaps
(report translated from Chinese language). China Cement Association. China, 2007.
Chen, L. H., Chen, K. S. and Tong, L. Y.; On the pyrolysis kinetics of scrap automotive tires.
Journal of Hazardous Materials B84, 43-55, 2001.
Chen, T., Wu, C., Liu, R., Fei, W. and Liu, S.; Effect of hot vapor filtration on the characterization
of bio-oil from rice husks with fast pyrolysis in a fluidized-bed reactor. Bioresource Technology,
102, 6178-6185, 2011.
Chinyama, M. P. M. and Lockwood, F. C.; Devolatilisation behaviour of shredded tyre chips in
combusting environment. Journal of the Energy Institute, Vol. 80, No. 3, 162-167, 2007.
CRH, Corporate Social Responsibility Report 2009, Ireland, 2009.
CSI (Cement Sustainability Initiative), Guidelines for alternative fuels selection and use, World
Business Council for Sustainable Development, Switzerland, 2005.
Directive 2010/75/EU. Available from: http://www.europa.eu.
ENS; Energistatistik, Energistyrelsen (Energy Statistics, the Energy Agency), Copenhagen,
Denmark, 2010. http://www.ens.dk.
European Commission; Reference Document on Best Available Techniques in the cement, lime and
magnesium oxide manufacturing industries. Seville, Spain, 2010. Available from
http://eippcb.jrc.es/reference.
51
Chapter 3 Alternative fuels in cement production
40
European Commision; European Pollutant Release and Transfer Register, 2011. Available online:
http://ec.europa.eu/environment/air/pollutants/stationary/eper/index.htm
FLSmidth, HOTDISC three years after. Highlights, April, 16-17, 2008.
FLSmidth, Latest generation HOTDISC successfully installed in Slovenia. Highlights, December,
26-27, 2009.
FLSmidth, HOTDISC Combustion Device. Brochure, FLSmidth, Valby, Denmark, 2011.
Gaur. S. and Reed, T. B.; Thermal Data for Natural and Synthetic Fuels, Marcel Dekker, Inc. New
York, USA, 1998. ISBN: 0-8247-0070-8.
Grasim, Sustainability Report 07-08. Grasim Industries Limited. Mumbai, India, 2008.
Hansen, J. P.; SO
2
Emissions from Cement Production. PhD Thesis, Technical University of
Denmark, Department of Chemical Engineering, 2003. ISBN 87-90142-96-9.
Hassan, E.-B. M., Steele, P. H. and Ingram, L.; Characterization of fast pyrolysis bio-oils produced
from pretreated pine wood, Appl. Biochem. Biotechnol., 154:182-192, 2009.
Heidelberg Cement; Annual Report 2010, Heidelberg, Germany, 2010.
Heidelberg Cement; Innovations in kiln gas bypass systems. Cement International, No. 2, Vol. 8,
61-66, 2010.
Hewlett, P. C. (Editor); Chemistry of Cement and Concrete; 4
th
edition, John Wiley & Sons Inc.,
New York, 1998. ISBN 0-340-56589-6.
Holcim, Corporate Sustainable Development Report 2009, Switzerland, 2009.
ICR Research; Sustained Efforts. International Cement Review, 74-78, August 2011.
Jensen, L. S.; NO
x
from cement production Reduction by primary measures. PhD Thesis,
Technical University of Denmark, Department of Chemical Engineering, 1999.
Jns, E. S. and stergaard, M. J. L.; Kiln shell corrosion. IEEE-IAS/PCA Cement Industry
Technical Conference, Vancouver, Canada, 343-359, April, 2001.
Karmaka, D. T. and Datta, A. B.; Generation of hydrogen rich gas through fluidized bed
gasification of biomass. Bioresource Technology, 102, 1907-1913, 2011.
Karstensen, K. H.; Formation, release and control of dioxins in cement kilns, Chemosphere, 70,
543-560, 2008.
Katsioti, M., Kasselouri, V., Ftikos, Ch. and Parissakis, G.; Control of chloride content and gas
temperature to prevent coating formation in a suspension preheater kiln. World Cement, 67-70,
October, 1995.
52
Chapter 3 Alternative fuels in cement production
41
Keefe, B. P. and Shenk, R. E.; An innovative solution for waste utilization, IEEE, 197-206, 2003.
Klauss, J. Burning cement clinker under reducing conditions in a rotary kiln, ZKG International,
132-144, Vol. 53, No. 3, 2000.
Kobyashi, N., Itaya, Y., Piao, G., Mori, S., Kondo, M., Hamai, M. and Yamaguchi, M; The
behaviour of flue gas from RDF combustion in a fluidized bed, Powder Technology, 151, 87-95,
2005.
Krennbauer, F.; Secondary fuels and their influence on the cement burning process; ZKG-
International, 5, 63-71, 2006.
Kntee, U., Zevenhoven, R., Backman, R. and Hupa, M.; Cement manufacturing using alternative
fuels and the advantages of process modelling, Fuel Processing Technology, 85, 293-301, 2004.
Lafarge, Annual Report 2010, Paris, France, 2010.
Larsen, M. B.; Alternative Fuels in Cement Production; PhD Thesis, Department of Chemical
Engineering, Technical University of Denmark, 2007. ISBN 978-87-91435-49-8.
Larsen, M. B.; Alternative fuels in cement production a new opportunity. Paper from FLSmidth.
Valby, Denmark, 2011.
Lee, V. K. C., Kwok, K. C. M., Cheung, W. H. and McKay, G.; Operation of a municipal solid
waste co-combustion pilot plant, Asia-Pac. J. Chem. Eng., 2, 631-639, 2007.
Lemarchand, D; Burning Issues, International Cement Review, 65-67, February 2000.
Locher, G. and Klein, H.; Modelling circulating sulfur, chlorine and alkali systems in the clinker
burning process; Part 2: Theory and discussion. Cement International, Vol. 7, 4, 64-75, 2009.
Nievoll, J., Jrg, S., Dsinger, K. and Corpus, J.; Studying ring formation. World Cement, 78-83,
December, 2009.
Norcini, M., Tassistro, N., Quaresima, R., Scoccia, G. and Volpe, R.; Use of chopped tyres as
auxiliary fuel in clinker kilns: Industrial experience at the Cagnano Amiterno cement works.
Proceedings of the 6th NCB international seminar on cement and building materials, 3, 1998.
Palmer, G.; Ring formation in cement kilns. World Cement, 538-543, December, 1990.
Panagiotou, T. and Levendis, Y.; A study on the combustion characteristics of PVC, Poly(stirene),
Poly(ethylene), and Poly(propylene) particles under high heating rates, Combustion and Flame, 99,
53-74, 1994.
Paone, P.; Heavy metals in the cement industry: A look at volatile cycles and simple mitigation
techniques. IEEE Cement Industry Technical Conference, 65-75, Miami, Florida, USA, 2008.
53
Chapter 3 Alternative fuels in cement production
42
PBL (The Netherlands Environmental Assessment Agency); Global CO2 emissions: increase
continued in 2007, Available from:
http://www.pbl.nl/en/publications/2008/GlobalCO2emissionsthrough2007.html, May 2011.
Peray, K. E.; The Rotary Cement Kiln; 2
nd
edition, Edward Arnold Ltd., Victoria, Australia, 1986.
ISBN 0 7131 3609X.
Pipilikaki, P., Katsiosi, M., Papageorgiou. D., Fragoulis, D. and Chaniotakis, E.; Use of tire-derived
fuel in clinker burning, Cement & Concrete Composites, 27, 843-847, 2005.
Potgieter, J. H. and Wirth, W.; An investigation into ash-ring formation in a rotary lime kiln. ZKG,
3, 166-171, 1996.
Prisciandaro, M, Mazziotti, G. and Vegli, F.; Effect of burning supplementary waste fuels on the
pollutant emissions by cement plants: A statistical analysis of process data, Resources,
Conservation and Recycling, 39, 161-184, 2003.
Rasmussen, M. H.; Low SO2 emission preheaters for cement production. PhD Thesis, Department
of Chemical and Biochemical Engineering, Technical University of Denmark, 2011.
Realff, M. J., Lemieux, P., Lucero, S., Mulholland, J. and Smith, P. B.; Characterization of transient
puff emissions from the burning of carpet waste charges in a rotary kiln combustor, IEEE
Conference, 212-228, 2005.
Saito, I., Sakae, K. Ogiri, T. and Ueda, Y.; Effective use of waste tyres by gasification in cement
plant, World Cement, 18, 264-269, 1987.
SCG; Sustainability Report 2010, Siam Cement Corporation. Bangkok, Thailand, 2010.
Scheuer, A.; Utilization of alternative fuels and raw materials in the cement industry, Cement
International, 1, 48-66, 2006.
Schmidthals, H.; The Pre-Combustion Chamber for secondary fuels development status of a new
technology, IEEE Conference, 207-218, 2003.
Schneider, M; Activity Report 2007-2009, Verein Deutscher Zementwerke, 2009,
http://www.vdz.de.
Seyler, C., Hellweg, S., Monteil, M and Hungerbhler, K.; Life cycle inventory for use of waste
solvent as fuel substitute in the cement industry, Int J LCA, 10 (2), 120-130, 2005.
Siam City Cement; Sustainable Development Report. Siam City Cement Public Co. Ltd., Bangkok,
Thailand, 2010.
Stenseng, M.; Pyrolysis and combustion of biomass; PhD Thesis, Department of Chemical
Engineering, Technical University of Denmark, 2001.
54
Chapter 3 Alternative fuels in cement production
43
Sutou, K., Harada, H. and Ueno, N.; Concentration is the key, International Cement Review, 36-41,
June 1999.
Sutou, K. and Yukio, K.; Using waste responsibly, International Cement Review, 2000.
Sylla, H.-M.; Investigations on the formation of rings in rotary cement kilns, ZKG, 499-508, 10,
1974.
Taiheiyo Cement Corporation; Corporate Social Responsibility Report 2010. Tokyo, Japan, 2010.
Tokheim, L.-A.; An alternative solution, World Cement, November 2005.
ThyssenKrupp, Forum Technische Mitteilungen, English version, Dec. 2002. Available from:
http://www.thyssenkrupp.com/documents/Publikationen/Techforum/Forum_12_02_eng.pdf.
Vrhegyi, G., Bobly, B.; Jakab, E. and Chen, H.; Thermogravimetric study of biomass pyrolysis
kinetics. A distributed activation energy model with prediction tests. Energy & Fuels, 25, 24-32,
2011.
VDZ, Environmental Data of the Cement Industry 2005. Available from: http://www.vdz-online.de.
VDZ, Environmental data of the German Cement Industry 2010. Available from: http://www.vdz-
online.de.
Wang, Z., Richter, H., Howard, J. B., Jordan, J., Carlson, J. And Levendis, Y.; Laboratory
investigation of the products of the incomplete combustion of waste plastics and techniques for their
minimization, Ind. Chem. Eng. Res., 43, 2873-2886, 2004.
Wanzura, F., Wendt, B.; Influence of the use of secondary materials on the levels of heavy metals in
Portland cement clinkers and cement. ZKG-International, 10, 53-60, 2003.
Wirthwein, R., Scharf, K.-F., Scur, P. and Drebelhoff, S.; Operating experience with a fluidized bed
gasifier using waste materials for lean gas making, ZKG International, 55, No. 1, 61-69, 2002.
Wirthwein, R.; Experiences in operating bypass systems at CEMEX Germany. ECRA Seminar,
Germany, September 2007.
Wurst, F. and Prey, T.; Dioxin emissions when using alternative fuels in the cement industry, ZKG
International, 56, No. 4, 74-77, 2003.
Zheng, Y.; Mercury removal from cement plants by sorbent injection upstream of pulse jet fabric
filter. PhD Thesis, Department of Chemical and Biochemical Engineering, Technical University of
Denmark, 2011.
55
Chapter 3 Alternative fuels in cement production
44
56
Chapter 4 Mixing of large and small particles in rotary kilns
45
4 Mixing of large and small particles in rotary kilns
This chapter seeks to provide a basic understanding of the mixing process of particles with
different sizes and densities in the material bed of rotary kilns. Through a literature study, the
present knowledge is described. Subsequently, sections about experimental work, results and
discussion will present the results obtained from raw material/ fuel particle mixing experiments
in a pilot-scale rotary kiln at ambient conditions. The chapter ends with a conclusion and
description of practical implications.
4.1 Introduction
Solid fuel particles entering a rotary kiln may be rapidly distributed into the raw material bed.
Depending on fuel and raw material particle characteristics as well as kiln operational
parameters, the solid fuels may be partly or fully buried in the raw material bed. The exact
position of the fuel particles has importance for the heat and mass transfer mechanism to the fuel
particles. For this reason it is important to obtain a basic understanding of the mixing process.
As a rotary kiln turns on its axis the solids bed inside is subjected to transverse motion. The bed
motion can take many forms, depending on variables such as rotational speed, kiln diameter, fill
degree, bed/wall friction and bed particle characteristics. The different bed motions - slipping,
slumping, rolling, cascading, cataracting and centrifuging, are illustrated in Figure 4-1. These six
characteristic flow patterns have great influence on the radial mixing, surface renewal, heat
transfer, dust generation and axial transport - for the continuous operation. For example, it is
widely recognized that the slipping mode, in which the bed slides against the inner kiln walls,
drastically reduces mixing of the solids and heat transfer. For conditions relevant for clinker
nodulization in cement rotary kilns, the characteristic bed motion is the rolling mode (Boateng,
2008). The main focus is therefore on the rolling mode. It should be noted however, that other
bed motions are possible, particularly in the material inlet end of the rotary kiln where the
partially calcined raw material particles enter with particle sizes typically in the order of 5-90 m
with diameter based average sizes around 20 m.
Figure 4-1: Bed behavior with Froude number tendency.
The surface of the rolling bed is approximately flat and inclined at an angle commonly described
as the dynamic angle oI repose, . The angle oI repose increases with increasing rotational speed
(Yang et al., 2003) and is reduced with increasing particle size (Rao et al., 1991). For cement raw
57
Chapter 4 Mixing of large and small particles in rotary kilns
46
materials the angle of repose is typically between 30-50 (Henein et al., 1983). The structure of
the bed comprises two layers, the active layer (surface) and the passive layer (sub-surface). This
is illustrated in Figure 4-2. In the passive layer the bed is closely packed and the particles rotate
with the kiln at fixed radius. Under these circumstances the possibility of particle mixing is quite
small (Ingram et al., 2005). When the particles reach the surface, they roll downwards and rejoin
the passive layer. In the active layer the particles are in motion and this is where mixing occurs.
The active layer thickness increases with increasing rotational speed and bed depth, because
more material is transported through the active layer as well as the passive layer per unit time.
The number of particles passing through the active layer increases also with increasing rotational
speed and decreasing particle size (Henein et al. 1983).
Figure 4-2: End view of rotary kiln showing active and passive layers as well as the angle of repose.
Henein et al. (1983) have performed extensive work to establish bed behavior diagrams for a
number of compounds. This type of bed behavior diagrams is useful to predict mode of bed
motion based on bed depth and Froude number (rotational speed). The Froude number is here
defined as:
2
2
2
60
n
R
R
Fr
g g
t
e
| |
|
\ .
= = (4.1)
where has the unit rad/s and n has the unit min
-1
. R is the drum radius and g is the gravitational
acceleration in m/s
2
. Mellmann (2001) investigated different operating regimes based on the
Froude number. He found that rolling beds typically occur for Froude numbers between 10
-4
< Fr
< 10
-2
for fill degrees higher than 10%. This result is in good correlation with Froude numbers
reported by Henein et al. (1983) from experiments in a rotary drum with limestone and sand.
Henein et al. found useful correlations for scaling-up. The bed behavior is not only characterized
by bed fill degree and Froude number for a given material. Particle size and shape must also be
considered. For particles with same shape and for kilns with the same diameter (D) and fill
degree (and same bed/wall and particle/particle friction), the bed motion from one particle size
(d
p
) to another can be described as (Henein et al., 1983):
58
Chapter 4 Mixing of large and small particles in rotary kilns
47
1/ 2 1/ 2
1 2
1 2
p p
D D
Fr Fr
d d
| | | |
=
| |
| |
\ . \ .
(4.2)
This correlation is also valid for scaling of bed behavior from one diameter to another, for the
same particle shape and size.
Cement rotary kilns are slightly inclined with the raw materials entering in the upper end and
flowing by gravity and kiln rotation to the lower end. A large fuel particle will spend some of its
time in the active layer, on the bed surface, and the remainder of the time in the passive layer
below the bed surface. The time that the fuel particle will be present in the active layer relative to
the time in the passive layer, depends on a number of parameters, such as bed fill degree,
rotational speed, raw material characteristics, fuel particle size, shape and density.
Although mixing occurs in both the axial and transverse direction, mixing in the transverse
direction is much more rapid (Sherrit et al., 2003). This is true for mixing in both industrial
rotary kilns as well as in horizontal, rotary drums.
The theoretical probability, P, of a large fuel particle to be visible in top of the bed rather than
being buried in the bed may be described as proportional to the volume occupied by the fuel
particle relative to the total available volume:
2
100%
4
fuel fuel fuel
fuel fuel
total
raw material
V V V
P
V V
V
V L D F
u u u
t
u u

+ +
(4.3)
where L and D is the kiln length and diameter, respectively, and F is the bed Iill degree. is the
sphericity which is a measure for the non-sphericity of the fuel particle, defined as:
of same volume
Surface of sphere
Surface of particle
u
| |
=
|
\ .
(4.4)
With this deIinition, 1 Ior spherical particles and 0 1 Ior all other particle shapes. The
sphericity is included because the shape of the particles is likely to affect the degree of visibility,
because for particles of the same volume, the external particle surface area increases when the
sphericity decreases. This may be illustrated with the three particle shapes shown in Figure 4-3.
All three particles have a volume of 1 cubic unit, but their external surface areas are quite
different depending on their shape. The surface area increases with decreasing sphericity and
thus the particle is more likely to be fully or partly visible in the bed.
59
Chapter 4 Mixing of large and small particles in rotary kilns

48

A = 4.86
= 1.00
A = 8.35
= 0.58
A = 10.50
= 0.46
d
10d
L
L/4
L/16

Figure 4-3: External surface areas and sphericities for three particles with the same volume.

Equation (4.3) includes the kiln dimensions, bed fill degree, volume and sphericity of the fuel
particle. The probability increases as the fuel volume increases. However, parameters such as
rotational speed, fuel density and raw material density are not included, even though they are
likely to have an effect. Attempts have been made to correlate experimental values for mixing
with the operating parameters and properties of granular material. Sherrit et al. (2003) provide a
review of such correlations. However, these correlations are based on mixing of particles with
the same or similar particle sizes. A similar correlation can be derived from experiments to yield
a correlation of the form:


2
100%
4
fuel
d
a
c raw material
fuel b
fuel
a
V
P Fr
V
D LF

u
t
u
| |
=
|
|
\ .
+
(4.5)

where a, b, c and d are experimentally determined fitting parameters. Fr is the Froude number
which includes the rotational speed.
raw material
and
fuel
is the density of raw material and fuel
particles, respectively.
Equations (4.3) and (4.5) are only valid for d
p, fuel
<< bed depth since fuel particles with all
dimensions exceeding the bed depth cannot be fully covered by the bed.

4.2 Literature study on mixing and segregation in rotary kilns and drums
Much research has been done to describe mixing of uniform particles in rotary kilns or rotary
drums (e.g. Woodle et al., 1993, Santomaso et al., 2005, Puyvelde, 2006 and Kwapinska et al.,
2006). Less information is available about the mixing of particles with different size, shape and
density. This is relevant when large alternative fuel particles are mixed with raw material
particles of much smaller particle size. The degree of mixing of these fuel particles is important
since their burnout behavior depends on mass and heat transfer. As it will be described in this
section, most of the current research about mixing of particles with different sizes has been made
with spheres of different materials, and typically with size ratios of less than 5.

When particles of different sizes and densities are mixed in the presence of a gravitational field,
segregation usually occurs. In theory, large low-density particles will drift towards the top of the
layer, followed lower down by smaller low-density and large high-density particles. The smallest
60
Chapter 4 Mixing of large and small particles in rotary kilns
49
high-density particles will collect in the bottom. The reason is that dense particles can push
lighter neighbour particles away and move downwards (Tanaka, 1971). Regarding particle size,
an explanation has been provided by Savage and Lun (1988): During mixing of several layers of
particles in a bed, void spaces between the particles are undergoing continual random changes. If
a void space is large enough, then a particle from the above layer can fall into it as the layers
move relative to each other. The probability of finding a void space that a small particle can fall
into is larger than the probability of finding a void space that a large particle can fall into. Thus,
this gravity-induced size-dependent void filling mechanism creates a tendency for small particles
to end at the bottom and large particles to end at the top.
Tallon and Davies (2008) have developed a method for in-situ measurement of mixing of
materials with different particle density or packing density. The principle is that a small cup is
mounted to the inner drum wall and is filled with material during rotation. The mass of material
in the cup is then measured by a load cell when the cup is above the free surface of the bed
particles. This gives a good indication of the mixing process.
Orpe and Khakhar (2001) studied the flow and segregation of 2 mm spheres of sand, glass and
steel in rotating cylinders. They found that particle surface roughness only had a very small
effect on the flow and segregation. However, similar experiments with 1-3 mm spheres of
stainless steel, mild steel and brass in rotating cylinders have shown that surface roughness can
affect segregation (Hajra and Khakhar 2004). Hajra and Khakhar furthermore reported that
segregation of large 2 or 3 mm spheres in smaller 1 or 1.5 mm spheres were independent of the
size of the larger spheres, but depended strongly on the size of the smaller spheres.
Clment et al. (1995) studied mixing of 3 mm steel spheres in 1.5 mm steel spheres in a rotating
cylinder and found that the smaller particles had a tendency to stay in the centre, while the larger
particles dwelled on the edges. This finding was confirmed by Cantelaube et al. (1995 and 1997)
who made experiments with glass particles of diameters from 6 to 20 mm. Cantelaube et al.
(1995) also reported that segregation of large particles appeared very quickly after the start of
drum rotation, typically in less than one drum revolution. Alonso et al. (1991) found that 1-4 mm
diameter spheres of different materials had a tendency to segregate with the smallest particles in
the core, but that particle size and density could mutually compensate to reduce segregation.
Thomas (2000) studied segregation of spherical glass spheres of variable diameter between 45
m and 7.5 mm. She found that there was a segregation of large particles at the bed surface only
if the size ratio between large and small particles were below 5. For size ratios above 5, the large
particles could be located anywhere in the bed depending on their size and mass. This was
explained by segregation being a result between various effects, counteracting each other. For
example, the geometrical void effect will push the large particles to the bed surface and the mass
effect will push it to the bottom of the bed by pushing away smaller particles.
For fuel particles in an industrial rotary kiln, the fuel particles will release volatiles during the
devolatilization and CO/CO
2
during the char oxidation, see chapter 5. The fuel particles may also
break into smaller particles during the combustion process. Thus, fuel particle density, shape and
size may change over time and this is likely to affect the mixing process.
61
Chapter 4 Mixing of large and small particles in rotary kilns
50
4.3 Experimental
An experimental study was carried out to investigate mixing of large solid fuel particles in a
rolling bed of cement raw materials. Key parameters such as fuel particle size and density were
studied as well as the parameters rotational speed and raw material fill degree.
4.3.1 Experimental set-up
A pilot-scale rotary kiln with inner diameter and length of 50 cm was constructed and used for
the experiments, see Figure 4-4. The kiln walls were made of a deck plate with a thickness of 5
mm, and a rough inner surface. The two end walls consisted of 600 mm diameter plexi glass
plates with a thickness of 5 mm, which were mounted with butterfly screws for easy removal
during loading and unloading of the pilot-scale rotary kiln. The pilot-scale rotary kiln was placed
horizontally on rollers connected to a motor with variable speed.
Figure 4-4: Sketch of pilot-scale rotary kiln used in the experiments.
62
Chapter 4 Mixing of large and small particles in rotary kilns
51
4.3.2 Bed materials and fuel characteristics
The experiments were all made with cement raw materials with physical and chemical properties
as shown in Table 4-1. The raw material has a particle size that is significantly smaller than that
of the fuel particles.
Chemical composition
CaCO
3
78.4 wt.-%
SiO
2
13.5 wt.-%
Fe
2
O
3
2.3 wt.-%
Al
2
O
3
3.3 wt.-%
Minor components 2.6 wt.-%
Bulk density 1,200 kg/m
3
Mean particle size 18 m
Sieve interval, m
0-5 21.4 wt.-%
5-10 13.3 wt.-%
10-20 19.0 wt.-%
20-30 12.5 wt.-%
30-60 18.9 wt.-%
60-90 8.2 wt.-%
90-400 6.6 wt.-%
Table 4-1: Composition and physical properties of cement raw material used in the experiments.
The 15 different fuel particles used in the experiments are presented in Table 4-2. These fuel
particles all have different shapes, sphericities, sizes, volumes, weights and solid densities, given
in the table per particle. These fuel particles were chosen because they are representative for the
fuel types that may be relevant to use in industrial rotary kilns. The sphericity is defined in
equation (4.4). As indicated by Table 4-2 some of the fuel particles have significantly larger
dimensions than the raw material, and the size ratio may be as high as 10
4
. The size ratio is,
however, realistic, since solid fuels used in industrial cement rotary kilns may also have
dimensions up to 1-2 m, whereas the cement raw material particle sizes generally are below 100
m.
63
Chapter 4 Mixing of large and small particles in rotary kilns
52
Sample
No. Particle type
Shape /
Sphericity
Size
[cm]
Volume
[cm
3
]
Weight
[g]
Density
[kg/m
3
]
1 Charcoal Oval / 0.95 5.5x5.5x3.0 56.8 45 792
2 Charcoal Oval /0.95 5.5x2.8x3.0 29.0 23 792
3 Charcoal Irregular / 0.89 2.8x2.8x3.0 10.1 8 792
4 Ceramic Spherical / 1.0 D=0.6 0.11 0.32 2829
5 Wood Cubic / 0.81 2x2x2 8.0 3.3 470
6 Wood Cylindric / 0.87 D=2.2, L=2 7.6 5.1 665
7 Wood Cylindric / 0.84 D=2.2, L=4 15.2 10 665
8 Wood Rectangular / 0.67 1.8x4.5x8.1 66.0 31 470
9 Wood Rectangular / 0.56 1.8x4.5x16.5 134.0 78 470
10 Tire rubber Rectangular / 0.78 2.5x2.0x4.0 20.0 22 1100
11 Tire rubber Rectangular / 0.75 2.5x5.0x6.0 75.0 82.5 1100
12 Tire rubber Rectangular / 0.59 2.5x8.0x17.0 340.0 374 1100
13 Plastic Cylindric / 0.73 D=1.3, L=5.3 7.0 2.4 341
14 Plastic Cylindric / 0.73 D=1.3, L=5.3 7.0 4.0 569
15 Plastic Cylindric / 0.73 D=1.3, L=5.3 7.0 7.7 1095
Table 4-2: Fuel particles used in the experiments. Size, volume and weight are per particle. The density is the solid
density.
4.3.3 Experimental procedure
The desired fill degree is obtained by filling the pilot-scale rotary kiln with the desired volume of
raw material. The fill degree is here defined as the percentage of the kiln cross sectional area that
is covered by raw material. A fixed number of fuel particles are placed on top of the raw
material, along the center line: 20 fuel particles were used in each experiment, except for the
experiments with the largest fuel particles, sample no. 9 and no. 12 in Table 4-2. For these
experiments with the two largest fuel particles, only 10 fuel particles were used. The rotation is
started with the desired rotational speed. The rotational speed used in the experiments is varied
between 3 and 16 rpm. It should here be noted that a rotational speed of 13 rpm in the
experimental set-up corresponds to approximately 5 rpm in an industrial rotary kiln with a 4 m
diameter, if the scaling correlations found by Henein et al. (1983) are used.
A video camera was used to record the mixing in order to document the experiments. At time
intervals of 30 seconds the rotation is stopped and the number of fuel particles visible on top of
the bed are counted. Then the rotation is started again. This start-stop cycle is repeated until
uniform mixing is achieved, or for at least 180 seconds. All fuel particles visible on the top of the
bed are included, even if a significant fraction of the fuel particle is buried in the bed.
64
Chapter 4 Mixing of large and small particles in rotary kilns
53
4.3.4 Assumptions and uncertainties
This section discusses the main differences between the pilot-scale rotary kiln and industrial
rotary kilns. The discussion presents assumptions that have to be made in order to account for
uncertainties when results may be transferred from the pilot-scale rotary kiln and to industrial
scale conditions.
1. The experiments are made at ambient temperature. It is a foregone conclusion that the
fuel particle/raw material mixing at ambient temperature is the same as at high-
temperature conditions. An uncertainty, however, is the surface properties of the particles
which may be quite different at ambient and high-temperature conditions. At high
temperatures the raw materials may stick to the fuel particles and e.g. affect sphericity
and flowability of the fuel particles.
2. The pilot-scale rotary kiln is not inclined. It is a foregone conclusion that the lack of
material transport in the axial direction has no significant effect on the overall mixing
process since mixing in the transverse plane is dominating (Sherrit, 2003).
3. It is a foregone conclusion that the dimensions of the pilot-scale rotary kiln are
sufficiently large to allow up-scale to industrial rotary kilns without introducing too much
uncertainty. This assumption is based on the fact that several researchers have used
similar rotary drums to simulate industrial conditions, and in many cases experimental
results have been successfully validated against published full-scale data (Henein et al.,
1983, and Mellmann, 2001).
4. It is a foregone conclusion that Henein et al.s (1983) scaling principles of rotational
speed for rolling beds can be applied for the conditions used in these experiments. This
assumption is confirmed by visual observations of bed motion prior to the experiments,
where the rolling motion is seen to be dominant.
5. It is a foregone conclusion that the bed-wall friction coefficient in the pilot-scale rotary
kiln is of comparable size to industrial kilns: The steel wall used in the pilot-scale rotary
kiln has a rough surface. In addition, the kiln walls are covered by a thin layer of raw
material that adheres to the surface. This is assumed to be comparable to an industrial
kiln where the wall is typically covered by a coating layer consisting of raw materials.
6. Under industrial conditions the fuel particles will release volatiles, CO and CO
2
during
the devolatilization and char oxidation. The fuel particles may also break into smaller
fragments. This induces a high degree of uncertainty on the mixing experiments made in
this study and the actual mixing process in industrial rotary kilns. It is, however, assumed
that the mixing experiments in this study will provide a fundamental understanding of the
mixing process under industrial conditions.
65
Chapter 4 Mixing of large and small particles in rotary kilns
54
4.4 Results and discussions
4.4.1 General observations
The mixing was generally observed to be a fast process that was completed in less than 30
seconds, or after a few kiln rotations. After this initial mixing, a steady state was reached were
the percentage of visible particles did not change significantly. This observation is in
correspondence with Cantelaube et al. (1995) who also reported that the mixing process appeared
very quickly after the start of drum rotation.
The rotational velocity is varied from 3 rpm to 16 rpm, corresponding to approximately 1 to 6
rpm in an industrial, 4 m diameter rotary kiln according to the scaling correlations found by
Henein et al. (1983). The bed behavior should be in the rolling regime under these conditions.
The rolling bed behavior was also observed to be dominant.
Henein et al. (1983) reported that the dynamic angle of repose should be between 30 and 50 for
cement raw materials in the rolling motion. He also reported that the angle of repose is
independent of fill degree and rotational speed when the rolling motion dominates. In the
experiments the angle of repose were generally measured to be between 30 and 40, and thus in
good correspondence with the findings of Henein et al. (1983).
All experiments have been repeated three times. The repeatability is generally good with
standard deviations of the experiments generally within 4-8%.
Based on the experimental data, optimal values for the fitting parameters a, b, c and d used in
equation (4.5) were determined. The fitting parameters were found by comparing equation (4.5)
with all experimental values and then modifying the equation until the best fit was found. The
fitting parameters are shown in Table 4-3 with uncertainties in parenthesis. The visibility
predicted by the correlation will be shown graphically in Figures 4-6 to 4-9. From the table it is
seen that the correlation actually only contains three fitting parameters since b = 1. In addition, it
should be noticed that the correlation may be further simplified by using c = 0. This will only
have a slight effect on the values predicted by the correlation and will reduce the number of
fitting parameters to two.
a b c d
1.5 0.2 1 0.2 -0.1 0.01 0.5 0.2
Table 4-3: Fitting parameters used in the correlation between percent visible particles and experimental data.
66
Chapter 4 Mixing of large and small particles in rotary kilns
55
4.4.2 Effect of fuel particle size and shape
The effect of particle size and shape on the mixing efficiency has been investigated for fill
degrees between 5% and 30% and rotational speeds between 3 rpm and 16 rpm. 12 samples with
different sizes and shapes were used in the experiments, see Table 4-2 for details. An example
for a fill degree of 10% and a rotational speed of 13 rpm is shown in Figure 4-5. It is observed
that particles, regardless of size and shape, decreases rapidly from 100% percent visibility to
significantly lower values. The mixing takes place within the first 30 seconds, corresponding to
1-8 kiln rotations. Subsequently, the particles are to a large degree covered by raw material and
only appear on top of the bed randomly.
It is noted that the large, rectangular samples number 9 and 12 are significantly more visible on
top of the bed: After 30 seconds approximately 47% and 58% of these particles are still visible,
respectively. And these percentages are more or less constant with time. By contrast, the smallest
particles which are the spherical ceramic particles denoted sample 4, drop from 100% to 0%
during the first 30 seconds. Subsequently, the small particles are not observed anymore on top of
the bed, during the 180 seconds that the experiment runs. The largest and most visible fuel
particles number 9 and 12 are also the least spherical particles, whereas the smallest and least
visible fuel particles are also the only truly spherical particles used in the experiments. These
results indicate that not only the fuel particle dimensions but also the fuel particle shape,
expressed as sphericity, are important for the visibility above the bed. This effect of sphericity is
further confirmed by comparing sample number 1 and 8 which have similar volumes of 57 cm
3
and 66 cm
3
, respectively, but quite different sphericities of 0.95 and 0.67. It is observed that the
least spherical particle is the most visible. However, the density of the two samples is not
identical and this may affect the visibility as well. In order to determine the effect of shape with
greater accuracy, it is necessary to conduct more experiments with fuel particles of similar
volumes and densities but different shapes.
The experiments have been repeated for different fill degrees and rotational speeds and the
tendency is the same under all conditions: Fuel particles with large dimensions have a higher
probability of being visible on top of the bed compared to smaller particles that are to a higher
degree buried in the bed. This geometric effect on the degree of visibility was also predicted by
equation (4.3) and (4.5).
67
Chapter 4 Mixing of large and small particles in rotary kilns
56
0 30 60 90 120 150 180
Time [sec.]
0
20
40
60
80
100
P
a
r
t
i
c
l
e
s
v
i
s
i
b
l
e
a
b
o
v
e
b
e
d
[
%
]
Sample No 1: Charcoal
Sample No. 2: Charcoal
Sample No. 3: Charcoal
Sample No. 4: Ceramic
Sample No. 5: Wood
Sample No. 6: Wood
Sample No. 7: Wood
Sample No. 8: Wood
Sample No. 9: Wood
Sample No. 10: Tire rubber
Sample No. 11: Tire rubber
Sample No. 12: Tire rubber
Figure 4-5: Example of raw data: Percent particles visible above bed as a function of time. 10% fill degree and 13
rpm.
Figure 4-6 shows the visibility of all the different fuel particles at steady state, sample no. 1 to
sample no. 12, as a function of the fuel particle volume. Steady state is here defined as the time
interval between 30 seconds and 180 seconds, where the percent visibility is relatively constant.
The small ceramic spheres, sample no. 4, had the smallest volume of 0.11 cm
3
per particle, and
had a steady state visibility of close to 0%. Several of the tested fuel particles had volumes
around 10 to 60 cm
3
, but different sphericities and densities. The steady state visibility is still
low, around 3-7%. For fuel particles with volumes larger than 100 cm
3
the percent visibility
increases significantly, up to 50% for the largest particle with a volume of 340 cm
3
. The standard
deviation varied from 0% to 8% during the experiments, and was observed to be greatest for the
largest fuel particles.
Figure 4-6 also shows the estimated probability for the particles to be visible based on equations
(4.3) and (4.5). Even though equation (4.3) does not take the effect of density and rotational
speed into account, a fairly good consistency is seen between the theoretical and experimental
values, for volumes up to around 100 cm
3
. But for the largest particles, equation (4.3) under-
estimate the probability: This is particularly observed for the particle with a volume of 134 cm
3
,
where the observed visibility is 43% but the theoretical probability is only 21%. The reason why
equation (4.3) fails to give a good estimation of the theoretical visibility for the largest particles
may be that it becomes increasingly difficult for the largest particles to be fully buried in the bed,
as their dimensions increases. Thus the degree of visibility will deviate more and more from the
theoretical predictions as the particle dimensions are increased.
Equation (4.5) which includes correlations for rotational speed and density gives a good
consistency between theoretical and experimental values for most of the particles. Fitting
parameters used are shown in Table 4-3. The greatest deviation is for the particles with volumes
of 57, 66 and 75 cm
3
which deviates 12%, 18% and 6%from the experimental values,
68
Chapter 4 Mixing of large and small particles in rotary kilns
57
respectively. One reason could be that these three particles have quite different volumes of 792
kg/m
3
, 470 kg/m
3
and 1100 kg/m
3
, respectively. It may be difficult to fit the model to give good
predictions of all combinations of densities, shapes, volumes and operational parameters.
0.1 1 10 100 1000
Particle volume [cm
3
]
0
20
40
60
P
a
r
t
i
c
l
e
s
v
i
s
i
b
l
e
a
b
o
v
e
b
e
d
[
%
]
Experimental data
Visibility predicted by equation (4.3)
Visibility predicted by equation (4.5)
Figure 4-6: Percent visible particles at steady state versus particle volume. 10% fill degree, 13 rpm.
The observations regarding fuel particle size and the degree of visibility above the bed is in good
correspondence with the theory about void filling presented by Savage and Lun (1988), which
also predicts that it becomes increasingly more difficult for a large particle to sink to the bottom
as the particle size increases. The observations may also be considered as a pure geometric effect
where increasing dimensions of the fuel particles imply a decreasing probability that the fuel
particles will be fully covered by the raw material bed.
69
Chapter 4 Mixing of large and small particles in rotary kilns
58
4.4.3 Effect of bed fill degree
The raw material bed fill degree has been varied for all samples in the following range: 5%,
10%, 15% and 20%. This range of fill degrees covers the situation in industrial cement rotary
kilns, where fill degrees typically are 8-15%. Figure 4-7 shows the effect of fill degree for tire
rubber particles of three different sizes and thus with three different volumes, as specified in
Table 4-2. For all three particle sizes, the percent visibility is observed to decrease with
increasing fill degrees. The large particles are around 80% visible at the 5% fill degree. At 10%
fill degree and 15% fill degree is the percentages around 50% and 40%, respectively.
0 5 10 15 20
Fill degree [%]
0
20
40
60
80
P
a
r
t
i
c
l
e
s
v
i
s
i
b
l
e
a
b
o
v
e
b
e
d
[
%
]
Particle surface area [cm
2
]
46
115
397
Figure 4-7: Percent visible tire rubber particles versus bed fill degree for three different particle sizes. Predicted
values are shown with solid lines and experimental values are shown with symbols. 10 rpm.
It was found that the fill degree is an important parameter for the percentage of fuel particles that
is present on top of the bed. Especially small particles will, regardless of shape, rapidly be buried
in the bed. Larger particles will, due to their proportions, only be partly buried in the bed
whereas part of the particles will be sticking out in the freeboard above the bed.
The degree of visibility predicted by equation (4.5), and with the fitting parameters shown in
Table 4-3, is shown with solid lines for each of the three particle sizes. The correlation provides
the right tendencies and generally fits the experimental data reasonably well. However, relatively
large deviations between model and experimental values of up to 20% are observed for the
medium sized tire rubber particles.
70
Chapter 4 Mixing of large and small particles in rotary kilns
59
4.4.4 Effect of bed rotational speed
Figure 4-8 shows the effect of rotational speed on the mixing of tire rubber particles, samples no.
10, 11 and 12 in Table 4-2. The rotational speed is expressed in terms of the dimensionless
Froude number. It is observed that the percentage of visible particles decreases when the Froude
number increases: Approximately 28% of the medium-sized particles are visible at the lowest
Froude number of 0.0024 (3 rpm), whereas only 9% are visible at the highest Froude number of
0.0702 (16 rpm). For the largest particles, 68% of the particles are visible at the lowest Froude
number and 48% are visible at the highest Froude number. The smallest particles, however, does
not show any particular dependence of Froude number. The same tendencies were also observed
for other fill degrees and other particles.
The degree of visibility predicted by equation (4.5) and with the fitting parameters from Table
4-3 is shown with solid lines for each of the three particle sizes. The correlation fits the
experimental data reasonably well with deviations generally less than 10%.
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Fr
0
20
40
60
80
P
a
r
t
i
c
l
e
s
v
i
s
i
b
l
e
a
b
o
v
e
b
e
d
[
%
]
Particle surface area [cm
2
]
46
115
397
Figure 4-8: Percent visible tire rubber particles versus Froude number for three different particle sizes. 10% fill
degree. Predicted values are shown with solid lines and experimental values are shown with symbols.
The effect of the rotational speed on percentage of visible particles can be explained by the fact
that a high rotational speed is equal to more bed revolutions and thus a more efficient mixing
process relative to a lower rotational speed. It should, however, be noted that the percent
visibility of particles seems to stabilize at a certain percentage for each rotational speed.
Thomas (2000) and Cantelaube et al. (1997) also studied the effect of rotational speed on the
segregation of particles up to 3 mm and 20 mm in diameter, respectively. They found no effect
of the rotational speed for these particle sizes, which are in correspondence with the observations
in this study, for the smallest particles.
71
Chapter 4 Mixing of large and small particles in rotary kilns
60
4.4.5 Effect of fuel density
The effect of particle density was studied by filling plastic containers with materials of different
weight. Thus, the outer dimensions of the particles were identical, but the densities were varied
from low, medium and high, see Table 4-2 for details. Figure 4-9 shows the percentage of visible
particles versus the raw material/fuel particle density ratio for fill degrees of 5%, 10% and 15%,
and rotational speed of 10 rpm. Standard deviations on the results presented in Figure 4-9 were
between 2% and 6%. It is observed that the greatest effect of particle density appears with the
5% fill degree: The low-density particles stabilize at a level around 18% visible particles, the
medium-density particles stabilizes at level around 11% visible particles, whereas the high-
density particles is only 4% visible. The same effect of particle density is also observed at 10%
and 15% fill degree.
Figure 4-9 also shows the values predicted from equation (4.5) and fitting parameters from Table
4-3, with solid lines and symbols. Equation (4.5) predicts the visibility excellent for the fill
degrees of 10% and 15%. But for the lowest fill degree of 5% there is a quite high deviation
between correlation and experimental values: The low-density fuel particles are around twice as
visible as predicted by equation (4.5). There is thus a poor relationship between the fitting
parameter for density, and some combinations of operational parameters.
1 1.5 2 2.5 3 3.5 4
Particle ratio [
Raw material
/
Fuel
]
0
4
8
12
16
20
P
a
r
t
i
c
l
e
s
v
i
s
i
b
l
e
a
b
o
v
e
b
e
d
[
%
]
5% fill degree
10% fill degree
15% fill degree
Figure 4-9: Percent particles visible above bed at steady state versus particle density. Estimated values are shown
with solid lines and symbols while experimental values are shown with symbols only. 10 rpm.
The effect of particle density on the percent visibility is in good correspondence with the push-
away theory presented by Tanaka (1971), which states that dense particles are more likely to
push away other particles and sink to the bottom.
72
Chapter 4 Mixing of large and small particles in rotary kilns
61
4.4.6 Effect of tumblers
The pilot-scale rotary kiln was equipped with four tumblers in order to study the effect on the
mixing process. The tumblers had a height of 80 mm and were placed symmetrically in the kiln,
see Figure 4-10. The tumblers are expected to improve the mixing process by lifting particles
from the passive layer up to 180 along the inner kiln wall before the particles falls down on the
active layer. The bed material fill degree will consequently be lowered, since some of the bed
material is lifted away from the bed. This is expected to expose more solid fuel particles to the
freeboard above the bed.
Figure 4-10: Four tumblers placed symmetrically in the pilot-scale rotary kiln.
A comparison of the percentage of visible tire rubber particles at 5%, 10% and 15% fill degrees
and 10 rpm with and without tumblers are shown in Figure 4-11. It is seen that the fuel particles
are generally more visible with the tumblers installed, typically with an improvement of 10-20%.
An exception is the medium-sized fuel particles at 5% fill degree: Here the visibility of the
particles is 58% without tumblers, which is higher than with tumblers, where the visibility was
50%. The standard deviations on the results with tumblers were generally higher than for the
results without tumblers, and were up to 17% in one case.
The degree of improved mixing with tumblers installed was found to increase with increasing
rotational speed.
73
Chapter 4 Mixing of large and small particles in rotary kilns
62
0 100 200 300 400
Particle surface area [cm
2
]
0
20
40
60
80
100
P
a
r
t
i
c
l
e
s
v
i
s
i
b
l
e
a
b
o
v
e
b
e
d
[
%
]
5% fill degree. No tumblers
10% fill degree. No tumblers
15% fill degree. No tumblers
5% fill degree. With tumblers
10% fill degree. With tumblers
15% fill degree. With tumblers
Figure 4-11: Comparison between percentage of visible tire rubber particles with and without tumblers at steady
state. 10 rpm.
Based on the experiments with tumblers it was found that tumblers generally improve the mixing
process. This is also in good correspondence with the fact that standard deviations on the results
with tumblers were generally higher than for the results without tumblers, because improved
mixing results in a more random distribution of the particles. Thus it becomes more difficult to
repeat the results of individual experiments, leading to a higher standard deviation.
During the experiments with the tumblers, it was clear that the bed behavior changed from
rolling to a more cataracting character: Raw materials and fuel particles were lifted up and fell
down again on top of the bed. The improved mixing caused by this new bed behavior is likely to
improve heat and oxygen transfer from the freeboard gas to the fuel particles in an industrial
rotary kiln. However, the tumblers will also result in a dustier atmosphere in the rotary kiln,
which may lead to more dust entrained in the flue gas leaving the rotary kiln.
74
Chapter 4 Mixing of large and small particles in rotary kilns
63
4.5 Practical implications
Equation (4.5) is applied to conditions of an industrial rotary kiln in order to estimate the effect
of fuel particle volume and sphericity on the percent visibility above the raw material bed.
Typical values for a cement rotary kiln are shown in Table 4-4.
Rotary kiln
Inner diameter, m 4.35
Total rotary kiln length, m 51
Material inlet section length, m 10
Rotational speed, min
-1
4
Angle of tilt, 3
Raw material flow, tpd (kg clinker/s) 3500 (40.51)
Mean raw material residence time, min 11
Mean raw material fill degree, % 11
Raw material bulk density, kg/m
3
1200
Table 4-4: Typical data for a modern cement rotary kiln (FLSmidth ROTAX 2 kiln).
Two selected fuels, tire rubber and wood are studied. The fuel densities will decrease rapidly
during the heating and devolatilization process in the material inlet where raw material
temperatures and gas temperatures are typically 900C and 1100C, respectively. It is a foregone
conclusion that the fuel char densities are representative during the mixing process. The char
density of tire rubber is assumed to be 340 kg/m
3
and the char density of wood is assumed to be
160 kg/m
3
.
Figure 4-12 shows the percent visibility of tire rubber char and wood char particles as a function
of the fuel particle volume and for three different fuel particle sphericities of 0.5, 0.75 and 1.0,
respectively. The volume of fuel particles is assumed to be distributed over the first 10 meters of
the material inlet end of the rotary kiln, where the combustion is assumed to take place. The
percent visibility is observed to increase linearly with the fuel particle volume. It is interesting to
note that the sphericity of the fuel particles have a great effect on the percent visibility: A pure
spherical Iuel particle ( 1.0) is least visible due to its compact shape which leads to a high
degree of fuel coverage by raw materials. As the fuel particle sphericity decreases, the percent
visibility increases due to the larger dimensions of the fuel particles. The increasing dimensions
of fuel particles of the same volume but decreasing sphericity are illustrated in Figure 4-3. The
tendency with sphericity and visibility is the same for tire rubber char and wood char but it is
observed that the wood char is considerably more visible than tire rubber char due to its lower
density.
75
Chapter 4 Mixing of large and small particles in rotary kilns
64
0 0.4 0.8 1.2 1.6
Particle volume [m
3
]
0
20
40
60
80
P
a
r
t
i
c
l
e
s
v
i
s
i
b
l
e
a
b
o
v
e
b
e
d
[
%
]
Tire rubber, Sphericity = 0.5
Tire rubber, Sphericity = 0.75
Tire rubber, Sphericity = 1.0
Wood, Sphericity = 0.5
Wood, Sphericity = 0.75
Wood, Sphericity = 1.0
Figure 4-12: Estimated degree of visibility of tire rubber char and wood char particles as a function of fuel particle
volume and sphericity. Estimated by equation (4.5).
An important practical implication for combustion processes in industrial rotary kilns can be
concluded to be the shape of the fuel particles, because low-sphericity fuel particles are expected
to be in better contact with the gas phase, which may improve the combustion process. Also the
fuel particle density will have an effect on the gas-fuel contact.
4.6 Conclusions for Chapter 4
During experiments in a pilot-scale rotary kiln distribution of large solid fuel particles in cement
raw materials were studied. The results were evaluated as the percentage of fuel particles visible
above the bed at specific time intervals.
The fuel particles were to a large degree buried in the cement raw materials within less than 30
seconds. Subsequently, the percentage of visible fuel particles stabilized at a specific level,
depending on fuel particle size. There was a clear tendency that large fuel particles were more
visible than smaller fuel particles. Fuel particle dimensions and sphericity are important
parameters for the degree of visibility: Large dimensions and low sphericity increases the degree
of visibility.
Low-density fuel particles were more present above the bed than high-density fuel particles
which were to a higher degree buried in the bed. The influence of density was particularly
pronounced at low fill degrees.
The bed fill degree is of major importance for the percentage of visible fuel particles in the bed.
Small bed fill degrees favor the number of fuel particles visible above the bed.
There is a tendency that an increase in rotational speed reduces the number of fuel particles
visible above the bed. However, the effect of rotational speed was not as significant as the effect
of bed fill degree.
76
Chapter 4 Mixing of large and small particles in rotary kilns
65
Installation of tumblers in the pilot-scale rotary kiln increased the percentage of fuel particles
visible above the bed, typically with a 10-20% improvement. However, the tumblers also led to a
more cataracting bed behavior which generates more dust entrained in the freeboard gas.
A correlation for the percentage of visible fuel particles and the operating parameters was
presented and compared with the experimental results. The correlation generally predicted the
degree of visibility reasonably well. However, the correlation failed to give good predictions in
some cases with low-density fuel particles.
The results indicate that large, solid fuel particles entering an industrial cement rotary kiln will
rapidly be fully or partly covered by cement raw materials. Thus, heat transfer by conduction
from the cement raw materials to the fuel particles must be expected to be a major heat transfer
mechanism rather than convection or radiation from the freeboard gas above the material bed.
Consequently, the temperature of the cement raw materials becomes of great importance for the
heating of the fuel particles. In addition, it must be expected that mass transfer of oxygen from
the freeboard gas to the fuel particles will be hindered by the cement raw materials covering the
fuel particles.
It was found that large fuel particles were in better contact with the freeboard than smaller fuel
particles of the same thickness. Thus, shredding of fuels to smaller particles is not necessarily
expected to improve the combustion process in the material inlet end of cement rotary kilns.
Since shredding is typically relatively energy-intensive and results in high wear on the shredding
equipment, there is reason to believe that it will be advantageous that solid fuels, e.g. whole tires,
undergo combustion in the material inlet end of cement rotary kiln as large particles rather than
small particles.
The results also indicated that the bed fill degree is of major importance for the gas-fuel contact,
whereas the rotational speed only showed a minor importance. Finally, installation of tumblers in
the material inlet end of cement rotary kilns may be useful to improve gas-fuel contact, thereby
improving the combustion process.
77
Chapter 4 Mixing of large and small particles in rotary kilns
66
4.7 References
Alonso, M., Satoh, M. and Miyanami, K.; Optimum combination of size ratio, density ratio and
concentration to minimize free surface segregation. Powder Technology, 68, 145-152, 1991.
Boateng, A. A.; Rotary kilns. Elsevier Inc., USA, 2008. ISBN: 978-0-7506-7877-3.
Cantelaube, F. and Bideau, D.; Radial segregation in a 2d drum: An experimental analysis.
Europhys. Lett., 30 (3), 133-138, 1995.
Cantelaube, F., Bideau, D. and Roux, S.; Kinetics of segregation of granular media in a two-
dimensional rotating drum. Powder Technology, 93, 1-11, 1997.
Clment, E., Rajchenbach, J. and Duran, J.; Mixing of a granular material in a bidimensional
rotating drum. Europhys. Lett., 30 (1), 7-12, 1995.
Hajra, S. K. and Khakhar, D. V.; Sensitivity of granular segregation of mixtures in quasi-two-
dimensional fluidized layers. Phys. Rev. E, 031304, 1-4, 2004.
Henein, H., Brimacombe, J. K. and Watkinson, A. P.; Experimental study of transverse bed
motion in rotary kilns, Metallurgical Transactions B, 14B, 191-205, 1983.
Ingram, A., Seville, J. P. K., Parker, D. J., Fan, X. and Forster, R. G.; Axial and radial dispersion
in rolling mode rotating drums, Powder Technology, 158, 76-91, 2005.
Kwapinska, M., Saage, G. and Tsotsas, E.; Mixing of particles in rotary drums - A comparison of
discrete element simulations with experimental results and penetration models for thermal
processes. Powder Technology, 161, 69-78, 2006.
Mellmann, J.; The transverse motion of solids in rotating cylinders - forms of motion and
transition behaviour, Powder Technology, 118, 251-270, 2001.
Orpe, A. V. and Khakhar, D. V.; Scaling relations for granular flow in quasi-two-dimensional
rotating cylinders. Phys. Rev. E, 64, 031302, 1-13, 2001.
Puyvelde, D. R. van; Comparison of discrete elemental modelling to experimental data regarding
mixing of solids in the transverse direction of a rotating kiln. Chem. Eng. Sci., 61, 4462-4465,
2006.
Rao, S. J, Khakhar, D. V. and Bhatia, S. K.; Axial transport of granular solids in horizontal
rotating cylinders: Part 2. Experiments in a non-flow system, Powder Technol. 67, 153162,
1991.
Santomaso, A., Olivi, M. and Canu, P.; Mixing kinetics of granular materials in drums operated
in rolling and cataracting regime. Powder Technology, 152, 41-51, 2005.
78
Chapter 4 Mixing of large and small particles in rotary kilns
67
Savage, S. B. and Lun, C. K. K.; Particle size segregation in inclined chute flow of dry
cohesionless granular solids. J. Fluid. Mech., 189, 311-335, 1988.
Sherritt, R. G., Chaouki, J., Mehrotra, A. K. and Behie, L. A.; Axial dispersion in the three-
dimensional mixing of particles, Chemical Engineering Science, 58, 401-415, 2003.
Tallon, S. and Davies, C. E.; In-situ monitoring of axial particle mixing in a rotating drum using
bulk density measurements. Powder Technology, 186, 22-30, 2008.
Tanaka, T.; Segregation models of solid mixtures composed of different densities and particle
sizes. Ind. Eng. Chem. Process Des. Develop., 10 (3), 332-340, 1971.
Thomas, N.; Reverse and intermediate segregation of large beads in dry granular media. Phys.
Rev. E, 62 (1), 961-974, 2000.
Woodle, G.R. and Munro, J. M.; Particle motion and mixing in a rotary kiln. Powder
Technology, 76, 241-245, 1993.
Yang, R. Y., Zou, R. P. and Yu, A. B.; Microdynamic analysis of particle flow in a horizontal
rotating drum, Powder Technol. 130, 138146, 2003.
79
Chapter 4 Mixing of large and small particles in rotary kilns
68
80
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles

69

5 Heat-up, devolatilization and combustion of large solid
fuel particles

This chapter seeks to provide insight into the heat-up, devolatilization and char combustion of large
solid fuel particles fired into the material inlet end of cement rotary kilns. Limited knowledge is
available about combustion of these large fuel particles, since focus has mainly been on suspension
firing of small coal or petcoke particles in the rotary kiln main burner or in the calciner.

In chapter 3 it was shown that solid alternative fuels in the cement industry comprises a long list of
different fuel types. The main emphasis in this chapter will be on tire-derived fuels (TDF) and wood
because TDF and wood is expensive to shred to smaller particle sizes, thereby making TDF and
wood attractive to utilize as coarse fuels fired into the material inlet of rotary kilns.

The experimental results are obtained from a pilot-scale, high-temperature rotary drum specifically
designed to simulate the process conditions in the material inlet end of industrial rotary kilns.

The chapter includes an introduction to solid fuel combustion and a literature study on
devolatilization and char combustion kinetics for large TDF and wood particles. The remaining part
of the chapter contains experimental investigations of large fuel particles and model analysis of
obtained results.


81
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
70
5.1 Introduction to solid fuel combustion
In this section a brief introduction to solid fuel combustion is presented. It is followed by a
description of heating, drying and devolatilization of solid fuel particles and finally a description of
the combustion of solid fuel char particles.
The combustion of a solid fuel may be divided into four stages: Heating, drying,
pyrolysis/devolatilization, gas and char combustion. These stages are illustrated in Figure 5-1. After
initial heating to around 100C, water evaporates and is transported away from the particle. The
next step is pyrolysis or devolatilization which is the release of volatiles without or with an
oxidizing atmosphere, respectively. In the following, the term devolatilization will be used, unless
reference is specifically made to a situation without oxygen being present. If sufficient oxygen is
available, the released volatiles will burn in a flame front at the surface of the fuel particle. The final
step is char combustion where the remaining fuel particle consisting almost exclusively of carbon
and ash inorganic minerals is converted into the ultimate residue, ash.
Figure 5-1: Heating, drying, pyrolysis/devolatilization and char combustion of a solid fuel particle (Hupa, 2008).
82
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
71
When a solid fuel particle is fired into the rotary kiln material inlet, heating mainly takes place by
conduction from the cement raw materials. In addition, convection and radiation from the gas phase
and kiln walls may contribute, depending on the actual position of the fuel particle in the rotary
kiln. This is illustrated in Figure 5-2 where fuel particles are shown under three different situations:
1) Fully covered by raw materials where heat transfer is by conduction from the hot raw
materials and where mass transfer of oxygen to the fuel particle is hindered by the raw
materials. Mass transfer of devolatilization and combustion products away from the fuel
particles may also be hindered by the raw materials in the bed.
2) Partly covered by raw materials where heat transfer is by all three mechanisms radiation,
convection and conduction. Mass transfer to and from the fuel particle may be partly
hindered by the raw materials in this position.
3) On top of the raw materials bed, where heat transfer is predominantly by radiation and
convection. However, even in this position a fraction of the heat transfer will be by
conduction from the raw materials since the fuel particle is always in physical contact with
the raw materials. Mass transfer between the fuel particle and freeboard gas is optimal in
this position.
The fraction of the particle surface that is covered by raw materials may be described as Y,
where Y = 1 corresponds to the situation where the fuel particle is completely covered by raw
materials and where Y = 0 corresponds to a situation where the entire fuel particle surface area
is exposed to the freeboard gas. Please refer to chapter 4 for a thorough discussion on
parameters affecting the Y-value. In chapter 4 was also presented the probability P, for fuel
particles to be visible above the bed, see equations (4.3) and (4.5). The relationship between Y
and P cannot be correlated by any simple expression because it is two fundamentally different
ways to describe the distribution of solid fuel particles in the bed.
Figure 5-2: Fuel particle at three different locations in the rotary kiln: Covered by raw materials, partly covered and
above the raw material bed.
83
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
72
5.1.1 Solid fuel conversion pathways
During devolatilization and char oxidation, a solid fuel particle can follow various conversion
pathways. The overall combustion kinetics of different solid fuels may be significantly affected by
the type of conversion pathway. Figure 5-3 shows examples of typical solid fuel conversion
pathways. Reaction path K illustrates the reshaping of virgin fuel particles which may occur during
heating. Reaction path A illustrates the direct devolatilization of gases from the solid fuel. Paths B
and C show formation of intermediate liquid and tar products, respectively, before being evaporated
to gas by path D. For devolatilization through reaction path I, the solid fuel particle shrinks during
devolatilization with or without simultaneous formation of smaller char particles. If sufficient
oxidizer is present, the gas produced through the devolatilization will be combusted in the gas phase
to the ultimate gaseous combustion products.
The char combustion may also follow different pathways. In pathway F the char fragments into
smaller char particles. In pathway G the char is converted as a shrinking particle and in pathway H,
the char is converted as a particle with constant size. Pathway H may be described by either an
unreacted shrinking core, H1, or a progressive conversion model, H2 (Levenspiel, 1999). In
pathway J, the agglomeration of char may create larger char particles that are subsequently
converted via pathway F, G or H.
It is important to understand which pathway a specific fuel particle follows during conversion,
because it can greatly influence the conversion time. Fuels that follows pathway I are for example
expected to reach full conversion much faster than fuel that follow pathway E. And if the char
cracks into smaller fragments, pathway F, the smaller particle sizes are expected to result in a faster
conversion relative to a larger char particle by pathway G, H and J.
84
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
73
Figure 5-3: Typical conversion pathways involved in the combustion process of a solid fuel particle (Larsen, 2007).
5.1.2 Heating of solid fuel particles
Upon heating to approximately 100C, moisture is evaporated and transported through the outer
particle volume and away from the particle surface. Thus the particle is dried. The heating of the
particle depends on the particle size: For small particles with diameters in the order of 10-100 m,
temperature during heat up can be assumed to be uniform throughout the particle. For larger
particles, however, internal temperature gradients may exist, and these need to be taken into
consideration (Agarwal et al., 1986).
An approximate definition of when the particle is isothermal and when it is non-isothermal is given
by the dimensionless Biot number, Bi:
2
p
p
hd
Bi
k
= (5.1)
Gas
Liquid
Tar
Oxidizer
Gaseous combustion products
Virgin solid
f uel particle(s)
Oxidizer Oxidizer Oxidizer
Oxidizer
Oxidizer Oxidizer
Ash
Oxidizer
Oxidizer
A
B
C
D
E
F
I
G
H1
H2
J
Converted via pathway F,G or H
K
Fuel with volatiles
Char without volatiles
Ash
85
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
74
The Biot number expresses the heat transfer to the particle relative to the rate at which heat is
transported into the particle. h is the heat transfer coefficient, d
p
is the particle diameter and k
p
is the
thermal conductivity of the particle. When Bi < 1, the particles may be considered to be isothermal,
while it is more complex for Bi > 1 where non-uniform temperatures inside the particle must be
taken into consideration.
5.1.3 Devolatilization of large fuel particles
Devolatilization of solid alternative fuels is important because alternative fuels typically contain a
relatively high amount of volatiles, see fuel analyses in chapter 3 section 3.2.3. As a consequence, a
high proportion of the heating value is released during the devolatilization.
For solid fuel particles where the char layer is retained during devolatilization, several physical and
chemical processes influence the devolatilization process, see Figure 5-4. The devolatilization
process is initiated by external heating when the particle surface temperature reaches the lower
devolatilization temperature, T
Vol1
. Heat is transported into the material by conduction through the
char, reacting and virgin layers. Radiative heat transfer may play a role in the char and reacting
layers. In the reacting layer, volatiles are released, leading to an outward mass transport through the
char and reacting layer. The gaseous devolatilization products typically consist of CO,
hydrocarbons (C
x
H
y
) and tar. The outward mass-flow may cool the hotter solid when the cooler
mass flows from the inner particle. Pyrolysis reactions may be either endothermic or exothermic,
and the latter may increase the intra-particle temperature locally and consequently conduct heat
both inwards and outwards. In an oxidizing atmosphere, the devolatilization products may ignite in
a flame front above the outer surface, enhancing the external heat transfer to the surface. This may
accelerate the overall devolatilization process depending on particle size. The ultimate
devolatilization products in an oxidizing atmosphere are CO
2
and H
2
O. The devolatilization can be
assumed to be finished when the centre of the fuel particle reaches the upper devolatilization
temperature, T
Vol2
.
86
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles

75


Figure 5-4: Radial cross section of partially devolatilized solid, cylindrical fuel particle (Larsen, 2007).

5.1.4 Char oxidation

After the devolatilization the remaining char is a product rich in carbon. Char oxidation may be
assumed to start when the devolatilization is finished, because the flame formed during the
combustion of the volatiles may effectively prevent oxygen transport to the surface of the char.
During the char conversion, oxygen is transported from the surrounding gas to the solid fuel outer
surface, and eventually also further into the particle matrix by pore diffusion. Oxygen is adsorbed to
the surface of the solid matrix, where it reacts with solid carbon and desorbs as gaseous combustion
products, CO and/or CO
2
. The products are transported to the surface and further away through the
boundary layer.

Large char particles often burn under diffusion limitations which means that the oxygen
concentration near the char particles is not constant. Thus the mass transfer rate of oxygen to the
char may become the dominant rate limiting parameter.

Char Reacting layer Virgin fuel
Mass Transport
Heat transport
Temperature
Solid density
Radius
Boundary
layer
87
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
76
5.2 Literature study on devolatilization and char combustion kinetics of large
particles
This section will provide an overview of devolatilization and char combustion of tire-derived fuels
and wood described in the literature. The focus is on large fuel particles, being of most relevance in
the material inlet end of cement rotary kilns.
5.2.1 Tire-derived fuels (TDF)
Devolatilization of large TDF particles have been studied by Bouvier et al. (1987), Yang et al.
(1995), Schmidthals (2001), Giddings et al. (2002), Larsen et al. (2006), Chinyama and Lockwood
(2007). In addition, several studies have been made with micro-TGA on small TDF samples to
determine reaction kinetics and composition of oil, gas and char (Conesa et al. (1998), Leung et al.
(1998) and Kyari et al. (2005)). Bouvier et al. (1987) studied the pyrolysis of cubic tire chips with
sizes from 5 to 20 mm at temperatures from 433C to 590C and under inert conditions. The
conversion times were approximately 100 s for a 5 mm cubic tire particle at 590C, while it was
around 400 s for a 20 mm cubic tire particle at 590C. Yang et al. (1995) studied the pyrolysis time
for a tire rubber cylinder with a diameter of 40 mm and a length of 60 mm under vacuum and at
510C. The pyrolysis time was approximately 50 minutes. Schmidthals (2001) found that TDF
particles in the sizes 10-20 mm x 50-70 mm pyrolysed within 10 minutes in air and within 20
minutes in nitrogen, both at 650C. Giddings et al. (2002) made a visual study of tire chip
devolatilization in a laboratory tube furnace at 900C. The tire chips had dimensions from 12 mm to
152 mm and masses from 2 to 30 g, which are not particularly well-defined particle sizes. But the
devolatilization time of a 7 g tire chip (no exact dimensions given) took approximately 120 s at
900C. The devolatilization time was reported to depend on the tire chip thickness. Larsen et al.
(2006) studied the devolatilization of cylindrical tire rubber particles with diameters from 7 to 22
mm and heights of 35 mm at temperatures up to 840C in an inert atmosphere. The reported
devolatilization times at 840C were 75 to 300 s when increasing the particle diameter from 7.5 to
22 mm. Chinyama and Lockwood (2007) studied the devolatilization times of tire rubber particles
with thicknesses in the range 6-12 mm and in the temperature interval from 700-1,000C. The
devolatilization times were found to be 30-100 seconds depending on thickness and temperature.
These results reported in literature about devolatilization of large TDF particles showclear
tendencies for devolatilization time to increase with increasing particle size and decrease with
increasing temperature and/or oxygen concentration.
Previous work on tire char combustion has been made with few mg of tire char particles in the 100-
500 m size in TGA reactors and drop tube reactors (Atal and Levendis, 1995, Masi et al., 1997,
Leung and Wang, 1998, Conesa et al., 1998, Larsen et al., 2007). Atal and Levendis (1995) studied
the combustion of tire char with particle size of around 500 m in air and at 1,200C. The char
conversion time was well below 1 s. Atal and Levendis also reported that tire char burned 2-4 times
faster than bituminous coal char. Masi et al. (1997) found that a TDF char particle with a diameter
of 100 m reached full conversion after 6 s at 5 vol.-% O
2
and 850C. According to Masi et al.
(1997), a medium rank coal char would require 100 s to reach full conversion under the same
conditions. Masi et al. (1997) also compared intrinsic TDF char reactivity with the intrinsic char
reactivity of RDF and biomass (Robinia Pseudoacacia) and found that the TDF char was most
reactive. Larsen et al. (2007) studied the combustion of tire char with particle size between 102 and
212 m at 10 vol.-% O
2
. The observed conversion times were from 9 s at 750C to 4 s at 850C.
Larsen et al. (2007) concluded that the TDF chars were very reactive at temperatures of 850C and
higher. As a consequence, intra-particle kinetics was concluded to be less important because the
88
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
77
reaction would take place at the outer particle surface. In addition, the ash layer formed on the
particle surface was observed to be very porous, and could easily be removed by slight mechanical
interaction. In the rotary kiln, the rotational movement of the kiln is assumed to remove the ash
layer rapidly, leaving an unconverted char particle.
5.2.2 Wood
Devolatilization of large wood particles has been studied by Winter et al. (1997), Di Blasi (2000),
de Diego et al. (2002), Di Blasi and Branca (2003) and Jand and Foscolo (2005). Winter et al.
(1997) measured temperature profiles in spherical 3-20 mm diameter beech wood particles at
oxygen partial pressures from 0 to 21 kPa and temperatures from 700C to 950C. The experiments
were made in a fluidized bed with silica sand as bed material. The devolatilization time of a 10 mm
diameter beech wood particle at 800C and oxygen partial pressure of 10 kPa was approximately 55
s. Di Blasi (2000) compared model predictions with measured values for beech wood particles in
the particle sizes 0.2-2 mm under fast pyrolysis conditions in a fluidized bed reactor. Conversion
times were reported to be 2-50 s at 527C. Devolatilization of non-spherical pine wood particles
with equivalent diameters, d
eff
, in the range 10 to 45 mm and in the temperature interval 650 to
850C has been investigated by de Diego et al. (2002). The reported devolatilization times for 10 to
30 mm diameter particles were for example from 30 s to 150 s at 850C. The devolatilization was
reported to be only slightly affected by a change in atmosphere from air to nitrogen. Di Blasi and
Branca (2003) studied temperature profiles in cylindrical beech wood particles with lengths of 20
mm and diameters from 2 to 10 mm. The experimental set-up was a fluidized bed with sand as bed
material and an inert atmosphere of nitrogen. Bed temperatures were from 439C to 834C. An
empirical correlation was suggested for the devolatilization time, and the devolatilization time for a
10 mm diameter beech wood particle at 834C was for example found to be approximately 40 s.
Jand and Foscolo (2005) studied the pyrolysis of spherical beech wood particles with diameters
from 5 to 20 mm and in the temperature interval 560-740C, in a fluidized bed gasifier with sand as
bed material. The devolatilization times at 740C were from 40 s to 140 s, depending on the particle
diameter.
The results reported in literature about devolatilization of large wood particles are as indicated
above typically based on experiments in fluidized bed reactors with sand as bed material and at
temperatures in the interval 400C-1,000C. The wood types are either beech wood or pine wood.
The general tendencies are that devolatilization times will increase with increasing particle size and
decrease with increasing temperatures. The effect of atmosphere seems to be of minor importance.
Most studies with wood char combustion are made with wood chars with particle sizes in the m
range. Examples from the literature are Janse et al. (1998), Senneca (2007) and Shen et al. (2009).
These studies typically use TGA experiments to obtain information about char intrinsic kinetics.
Jand and Foscolo (2005) studied the char combustion of large spherical beech wood char particles
with diameters from 5 to 20 mm at 740C and in air in a fluidized bed reactor. The char oxidation
times were found to be approximately 150 s for a 5 mm diameter char particle and 500 s for a 20
mm diameter char particle. Mermoud et al. (2006) studied the steam gasification of large spherical
beech wood particles with initial diameters of 10 to 30 mm in a macro-TG apparatus. The
temperatures were 830-1,030C and the steam partial pressures were 0.1-0.4 atm H
2
O. The char
conversion time was found to depend strongly on particle size, steam partial pressure and
temperature. From this study it also appears that steam gasification of wood char is relatively slow
compared to char oxidation in an oxidative atmosphere: For example, the conversion time for a 20
mm diameter particle at 930C and 0.2 atm H
2
O was 2,500 s whereas the char oxidation time for a
89
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
78
20 mm diameter particle at 740C and 21 vol.-% O
2
was 500 s, according to Jand and Foscolo
(2005). However, the results reported in literature do indicate that carbon conversion from fuel
chars in rotary kilns may not only depend on oxygen concentration, temperature and particle size
but also on H
2
O concentration in the bulk gas.
5.3 Experimental
In order to study the combustion of solid fuel particles under conditions similar to those in the
material inlet end of an industrial rotary kiln, a new experimental set-up has been designed and
constructed. This section will describe the requirements to this experimental set-up as well as the
practical considerations and assumptions that have been made.
5.3.1 Pre-experimental considerations
It is of interest to be able to investigate parameters such as rotational speed and bed fill degree. A
systematic investigation of these parameters coupled with gas composition, gas velocity,
temperature, characteristics of the fuels and raw materials may provide new knowledge. Two
specific topics should be investigated with the experimental set-up:
1. Systematic investigation to predict sensitivity of sulfur chemistry towards solid alternative
fuels: This is the subject of chapter 6.
2. Systematic investigation to predict combustion behavior for selected alternative fuels as a
function of key process parameters.
The information gained through these systematic investigations may help to identify critical
operational limitations regarding use of alternative fuels in the material inlet end, e.g. optimal fuel
types and amounts, conversion times and extent of sulfur release from the raw materials.
The experimental set-up must be able to simulate the process conditions in the material inlet end of
an industrial rotary kiln. The flue gas temperature in the material inlet end of cement rotary kilns is
typically 1,000-1,200C and the temperature of the fully or partly calcined raw materials from the
bottom stage cyclone is typically 850-890C. The solid fuels have not been heated prior to being
fired into the rotary kiln. It is thus required to be able to feed solid fuel at ambient temperature into
raw materials which have been preheated to 800-900C. The raw materials must be continuously
rotated in order to have the raw material bed in a predominantly rolling motion.
In an industrial rotary kiln, there is a constant heat transfer from the hot flue gas to the raw material
bed and rotary kiln walls due to the temperature gradients. Temperature profiles at specific
positions along the axial rotary kiln length can be maintained due to the continuous flow of gases
and raw materials. In a pilot-scale experimental set-up made for batch experiments with addition of
one batch of fuels to preheated raw materials, a stable temperature profile cannot be maintained if
gas of one specific temperature passes a bed of raw materials of another temperature: The
temperature gradient will gradually reduce until both gas and raw material bed reach the same
temperature. If gas and bed temperatures change during the experiment, it becomes uncertain what
the actual average temperature was during the experiment. In order to eliminate this uncertainty it
was decided to make the experiments under conditions where gas and raw material bed had the
same temperature. This approach also has the advantage that no heat transfer takes place between
gas, walls and bed, if there are uniform temperatures at all positions. Thus it is possible to neglect
90
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles

79

considerations about whether heat transfer should be by radiation and convection from the gas to
the bed or by radiation and conduction from the walls to the bed.

5.3.2 Experimental set-up
After careful consideration it was decided to construct a high-temperature rotary drum based on a
commercially available chamber furnace. The chamber furnace was supplied by Nabertherm and
can reach a maximum operating temperature of 1,300C. It is electrically heated by heating
elements embedded in two side walls and the floor. The chamber furnace was modified by drilling
holes in the rear wall and door for positioning of a steel drum, as seen in Figure 5-5. The steel drum
is made of the steel type Sandvik MA253 which can be used for temperatures up to 1,150C. The
drum is supported by a steel tube on rollers which passes through the rear wall hole in the chamber
furnace. This steel tube is connected to a motor by means of a roller chain, in order to rotate the
drum. The part of the steel tube that is outside the chamber furnace is water cooled in order to
protect the rollers, roller chain and gas seals against the high temperatures.





Figure 5-5: High-temperature rotary drum. Left: Side view. Right: Top view.

Gas can be transported to the chamber furnace through a hole in the roof. The furnace has been
designed to achieve sufficient heating of reactant gas before entering the rotary drum, which was
also verified by temperature measurements. The gas is transported into the rotating steel drum due
to an externally placed gas pump that pumps the gas through the rotating drum and steel tube
whereafter the gas exits the reactor for subsequent cooling and analysis.
The chamber furnace door is equipped with a window for visual inspection, two holes for
thermocouples and a centrally placed water-cooled tube for solid fuel addition. The solid fuels are
placed in a sample container which can be pushed into the hot rotary drum or pulled out to the
water-cooled tube.
The experimental set-up is shown in Figure 5-6. The gas supply can be up to 500 NL/min of air and
nitrogen. The gas is transported to the rotary drum reactor during the experiments, but may also be
bypassed during calibration of flow controllers, leakage tests, etc.
The temperature is measured at three different locations in the rotary drum reactor: At the rear wall
in the chamber furnace (T1), in the center of the rotary drum (T2) and at the door just in front of the
rotary drum (T3). A pump transports the flue gas out of the rotary drum reactor. Before passing the
pump, the flue gas is cooled in a heat exchanger and soot particles are captured in a filter. After
having passed the pump, the flue gas is either sent directly to the stack, or to the gas analyzers for
measurement of O
2
, CO, CO
2
, SO
2
and NO. The fraction of the flue gas that are transported to the
91
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles

80

analyzers will pass another two filters, a gas cooler for condensation of water and a sample gas
pump before reaching the gas analyzers.


Figure 5-6: Sketch of the high-temperature rotary drum experimental set-up.

The key specifications for the experimental set-up are presented in Table 5-1. The maximum
operating temperature should not exceed 1,000C, because the drum steel strength decreases
drastically at high temperatures. The volumetric raw material fill degree may be up to 17%. For
practical reasons, however, the fill degree should be lower, because some of the raw materials will
drop out of the rotary drum if the rotary drum is filled to the edge.

Parameter Value
Operating temperature _ 1,000C
Rotational speed 0-20 min
-1
Pressure Atmospheric, 25 mbar
Gas flow _ 500 NL/min
Oxygen concentration _ 21% O
2
Load capacity _ 6 L
Fill degree _ 17%
Table 5-1: Key specifications for high-temperature drum experimental set-up.

The amount of solid fuels per experiment depends strongly of the type of fuel: The fuel amount
increases as the volatile content in the fuel decreases. For a fuel consisting of pure volatiles, the fuel
amount is limited to the order of 5 g per experiment due to the rapid devolatilization which
increases the pressure in the experimental set-up and heavy soot deposition in the filters due to
incomplete fuel oxidation in the furnace. The solid fuel volume can be increased as the content of
volatiles decreases. For a fuel consisting of pure char, the fuel amount per experiment may be
significantly higher, e.g. in the order of 50 g, since the fuel char oxidation is controlled by a
relatively slow mass transfer of oxygen to the fuel particle.




92
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
81
5.3.3 Fuel samples
Proximate and ultimate analyses as well as lower heating values for the tested fuel samples are
shown in Table 5-2. Tire rubber and pine wood are cut into cylindrical and rectangular shapes,
respectively. The main focus has been on single particles of tire rubber and pine wood, but
experiments have also been conducted with polypropylene and petcoke. Polypropylene and petcoke
has been chosen as additional reference fuels due to their contents of volatiles and char representing
limiting cases for solid fuels. The experiments with polypropylene and petcoke were not with single
particles but instead with a batch of several small particles. The polypropylene was in the form of
rectangular flakes with a thickness of 1 mm. The petcoke particles were approximately spherical
with average diameters - 1 mm.
Proximate analysis Ultimate analysis LHV
VM FC Ash C H N S MJ/kg
Tire rubber 64.6 32.6 2.8 87.4 7.1 0.3 1.2 36.9
Pine wood 75.3 24.5 0.2 38.9 5.2 0.1 - 16.2
Polypropylene 97.5 0.0 2.5 83.0 14.0 0.0 - 44.5
Petcoke 13.4 85.1 1.5 87.3 3.7 1.5 4.7 34.0
Table 5-2: Fuel analyses and lower heating values (LHV) for solid fuels used in the experiments. Units are in wt.-% as
received and MJ/kg.
5.3.4 Experimental method
Raw materials are placed in the drum to obtain the desired volumetric fill degree. The reactor door
is closed, and the reactor is heated to the desired temperature. The drum always rotates when the
reactor is heated, and keeps the raw materials in a rolling motion, with an angle of repose of
approximately 30-40. It is the thermocouple T2 in the center of the rotary drum that is used to
determine the temperature prior to and during the experiments. The fuel sample is placed in the
sample boat, and the sample boat is then positioned in the water-cooled tube. The water-cooled tube
is closed in the end, in order to obtain a controlled atmosphere inside the rotary drum reactor. The
gas to the reactor is adjusted to the desired flow and oxygen concentration. The temperature at T2
and oxygen concentration at the O
2
analyzer is monitored. When a stable temperature and oxygen
concentration are reached, the fuel sample boat is pushed into the rotary drum, turned 180 and
pulled out in the water-cooled tube again. The fuel sample will thus drop into the preheated raw
materials and immediately be heated, followed by devolatilization and char oxidation. Temperatures
and flue gas compositions are continuously logged during the experiments with 1 s intervals.
An example from a typical experiment with combustion of one tire rubber cylinder is shown in
Figure 5-7. The oxygen concentration is observed to decrease from 10 towards 7 vol.-% when the
devolatilization starts. At the same time, the CO concentration increases from 0 ppm to
approximately 450 ppm, and the CO
2
concentration increases from 0 to 2.5 vol.-% during
devolatilization. The char oxidation is assumed to be finished when the CO
2
-concentration becomes
lower than 0.04 vol.-%. This assumption is in good agreement with the visual observations during
the experiment where red-glowing char particles can be observed in the bed until the CO
2
-
concentration reaches values below around 0.04 vol.-% CO
2
.
93
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
82
0 100 200 300 400
Time [s]
0
2
4
6
8
10
O
2
a
n
d
C
O
2
c
o
n
c
e
n
t
r
a
t
i
o
n
[
V
o
l
.
-
%
]
0
100
200
300
400
500
C
O
c
o
n
c
e
n
t
r
a
t
i
o
n
[
p
p
m
V
]
O2
CO2
CO
Figure 5-7: Example of experiment with combustion of a tire rubber cylinder. 900C and 10 vol.-% O
2
. 5% volumetric
fill, coarse sand, 6 rpm, 100 NL/min. Particle dimensions: D = 9 mm and L = 25 mm.
The degree of fuel conversion against time is derived, assuming that the fuel conversion is
proportional to the carbon conversion, by integration of the concentration profiles of CO
2
and CO:
2
2
0
0
( )
t
CO CO
CO CO
y y dt
X t
y y dt

+
=
+
}
}
(5.2)
5.3.5 Repeatability and uncertainties
The repeatability of the experiments is generally found to be acceptable. Nearly all experiments
were repeated at least three times to ensure the repeatability. However, individual experiments have
only been made once or twice if they were not part of the initial plan for experiments. The filters in
the experimental set-up have been cleaned regularly in order to avoid increases in pressure drop
over the system, which were observed to affect the repeatability of individual experiments. Carbon
mass balances were also performed regularly to test if all released carbon could be detected as CO
or CO
2
.
A sensitivity analysis was conducted to estimate the uncertainty on the degree of fuel conversion,
see Appendix A. The sensitivity analysis indicates an overall relative uncertainty on the estimated
degree of fuel conversion in the order of 6%.
94
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
83
5.4 Results and discussion
In this section, the experimental results are presented and analyzed. The main focus with the
analysis is to identify the main rate limiting parameters for the devolatilization as well as char
oxidation.
5.4.1 General observations
During the experiments it was possible to monitor the devolatilization and char oxidation process
visually through the window. This gave a good impression of the main reaction pathways for the
different fuels. For all the fuels, a flame front surrounding the fuel particles was observed almost
immediately after insertion into the rotary drum, indicating the start of the devolatilization process.
This is shown in Figure 5-8, left, which shows a tire rubber particle shortly after insertion into the
rotary drum.
Figure 5-8: Left: Tire rubber particle during devolatilization. Right: Tire rubber char fragments during char oxidation.
The tire rubber particles were observed to predominantly keep the particle shape under the
devolatilization but with some degree of fragmentation. After the devolatilization, the tire char
fragmented into several smaller char particles, as shown in Figure 5-8, right. This observation
indicates that a model for tire char oxidation should not be based on initial fuel particle size; it may
be based on the initial fuel mass and fixed carbon content instead. In this way, uncertainties about
actual particle sizes during the char combustion are avoided.
The pine wood and petcoke particles kept their shape during the devolatilization process and, to a
large extent, also during the char oxidation process. This observation indicates that an unreacted
shrinking core model could be suitable to describe the pine wood char and petcoke char oxidations,
see the conversion pathways in Figure 5-3.
The polypropylene plastic particles melted and behaved like a boiling liquid during the
devolatilization process.
Most of the experiments have been made with coarse sand as the raw material. This raw material
was chosen because 1) it is largely inert and 2) gives the desired rolling motion of the bed. It was
attempted to use real cement raw material, but without success due to the sticky nature of the raw
material which partly resulted in a thick coating layer on the inner drum wall and partly in clinker
nodulization.
95
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
84
5.4.2 Effect of fuel sample mass
The effect of fuel sample mass has been investigated with tire rubber, pine wood and petcoke.
Figure 5-9 left shows the conversion time versus mass for tire rubber cylinders with dimensions D =
9 mm and L = 25 mm and roughly spherical tire rubber granulate particles with D = 2 mm. The
masses are in the interval 1.8 g to 15 g and the results shown are made at 900C, 10 vol.-% O
2
and
in an empty rotary drum. It is observed that the conversion time increases linearly with mass for the
tire cylinders as well as for the tire granulate. It is furthermore observed that there are no particular
differences in conversion time for the tire cylinders and tire granulates when the mass is the same.
Thus the initial fuel particle size of the tire rubber does not have any particular effect on the
conversion time. This result is in good correspondence with the visual observation that the tire
rubber rapidly cracks into several small fragments after the devolatilization so conversion time will
be a function of overall tire rubber mass rather than tire rubber particle size.
The results are quite different with pine wood and petcoke where the conversion time was observed
to be a function of both particle size and sample mass, indicating that these fuels follow another
conversion pathway than tire rubber, as illustrated in Figure 5-3.
0 4 8 12 16
Mass of tire rubber [g]
0
1
2
3
4
5
C
o
n
v
e
r
s
i
o
n
t
i
m
e
[
m
i
n
]
Tire granulate (D = 2 mm)
Tire cylinder (D = 9 mm, L = 25 mm)
0 10 20 30
Mass of pet coke [g]
0
20
40
60
80
100
C
o
n
v
e
r
s
i
o
n
t
i
m
e
[
m
i
n
]
5 vol.-% O2
10 vol.-% O2
21 vol.-% O2
Figure 5-9: Left: Conversion time versus mass for tire rubber granulate and cylinders. 10 vol.-% O
2
. Right: Conversion
time versus mass for petcoke at different oxygen concentrations. Petcoke particle size = 2-4 mm. Empty rotary drum,
900C, 6 rpm, 100 NL/min. Conversion times evaluated at 80% carbon conversion.
Figure 5-9 right shows the conversion time for petcoke particles in the size interval 2-4 mm, and at
masses from 1 g to 30 g. The conversion time is observed to increase linearly with sample mass and
is observed to be significantly longer than for tire rubber. It is also observed that the conversion
time is roughly halved when the oxygen concentration increases from 5 vol.-% to 10 vol.-% and
from 10 vol.-% to 21 vol.-%. This illustrates the strong dependency of the bulk oxidizer
concentration for petcoke which predominantly consists of char. The effect of oxygen concentration
for pine wood and tire rubber will be shown in section 5.4.6.
It should be noted that some of the experimental data for tire rubber and petcoke combustion shown
in Figure 5-9 has been obtained by project students Yu Jiang and Dawid Jan Bialas, respectively.
96
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
85
5.4.3 Effect of fuel particle size and shape
The effect of fuel particle size has been studied with cylindrical tire rubber particles and rectangular
wood particles. The rectangular wood particles are also used to study the effect of fuel particle
shape. Figure 5-10 left shows the conversion curves for three tire rubber cylinders of different
diameters (and different masses) at 900C and 10 vol.-% oxygen. The overall conversion time is
observed to increase when the particle diameter (and mass) increases. It should be noted that the
devolatilization time also increases when the particle diameter increases. However, based on the
study of fuel sample mass shown in Figure 5-9 it must be concluded that the increase in overall tire
rubber conversion time is due to the increase in mass rather than the increase in particle diameter,
since the tire char rapidly fragment into several smaller char particles.
0 200 400 600 800
Time [s]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
Dp = 9 mm (1.8 g)
Dp = 12 mm (4.4 g)
Dp = 16 mm (6.3 g)
0 4 8 12 16 20
Time [min]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
30 x 15 x 10 mm (1.8 g)
30 x 30 x 10 mm (3.5 g)
60 x 30 x 10 mm (7.0 g)
120 x 15 x 10 mm (7.4 g)
30 x 30 x 20 mm (7.2 g)
Figure 5-10: Left: Carbon conversion for three different tire rubber cylinders. L
Cylinder
= 25 mm. Right: Carbon
conversion for pine wood particles of different sizes. T = 900C, 5% fill, coarse sand, 10 vol.-% O
2
, 6 rpm, 100
NL/min.
Figure 5-10 right shows the carbon conversion versus time for five pine wood particles with
different masses, sizes and shapes. Four of the particles have a thickness of 10 mm while the fifth
particle has a thickness of 20 mm. It is observed that the devolatilization time is approximately 1
minute for the four 10 mm thick particles whereas the devolatilization time is around 2 minutes for
the 20 mm particle the exact devolatilization times are shown in the model analysis, section 5.5.2.
The char oxidation times are generally observed to increase when the mass of char increases, but it
is also observed that the shape influences the char oxidation time. Three of the particles have
approximately the same mass of 7 g, but different char oxidation times: The longest of the 10 mm
thick particles, with dimensions 120 x 15 x 10 mm, is fully combusted after approximately 11
minutes while the shorter, more compact particles with dimensions 60 x 30 x 10 mm and 30 x 30 x
20 requires approximately 16 and 18 minutes for full conversion, respectively. Regarding particles
with similar mass but different shapes, the reason for the difference in conversion times may be
explained by the external surface area of the particles, which is largest for the particle that reaches
full conversion first.
Overall, the results indicate that particle thickness and external surface area have an effect on both
the devolatilization time and the char combustion time.
97
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
86
5.4.4 Effect of raw material particle size
Four different raw material types have been tested, see Table 5-3. The first raw material type was
authentic raw meal from a cement plant. However, this raw material could not be used for the
experiments because of its sticky characteristics during heating: The raw meal adhered to a high
degree to the rotary drum inner walls and formed nodules during the experiments. In addition, the
raw meal did not roll in the drum but was instead in a slumping motion, see Figure 4-1.
The fine sand did only to a small degree adhere to the inner rotary drum walls, but it was also
predominantly in a slumping motion rather than in the desired rolling motion.
The medium and coarse sand did not adhere to the rotary drum inner walls and were in the desired
rolling motion, making these two raw material types suitable for the experiments.
Raw material type Particle size distribution Average particle size (d50%)
Raw meal 2-140 m 18 m
Fine sand 2-140 m 31 m
Medium sand 90-500 m 250 m
Coarse sand 400-800 m 600 m
Table 5-3: Raw materials tested in the experiments.
Figure 5-11 shows the conversion versus time as a function of the raw material type for one pine
wood particle with dimensions 60 x 30 x 10 mm and 10 g of small petcoke particles with average
diameters of 1 mm. It is observed that the time for pine wood devolatilization is identical regardless
of the raw material. The char oxidation time for the medium sand and the coarse sand are practically
the same, approximately 17 minutes for pine wood and 40 minutes for petcoke. But the time for
char oxidation is significantly longer for the fine sand than for the medium and coarse sand,
approximately 35 minutes for pine wood and 100 minutes for petcoke. The reason is believed to be
that the fine sand follows a slumping bed motion whereas the medium and coarse sand follows a
rolling bed motion. In the slumping motion, mixing is less efficient than in the rolling motion,
which means that covered fuel particles will be in poor contact with oxygen from the freeboard gas.
Another reason may be that diffusion of oxygen through the bed to the fuel particle is slower when
the bed material is fine sand relative to coarse sand.
The results for tire rubber conversion are not shown, but the tendency was the same: The
conversion time were significantly longer with the fine sand as bed material relative to tire
conversion times in the medium and coarse sand.
98
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
87
0 20 40 60 80 100
Time [min]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
Wood in fine sand
Wood in medium sand
Wood in coarse sand
Pet coke in fine sand
Pet coke in medium sand
Pet coke in coarse sand
Figure 5-11: Carbon conversion for pine wood and 10 g petcoke particles in different raw materials. Wood particle
dimensions: 60 x 30 x 10 mm. Petcoke average particle diameter: 1 mm T = 900C, 5% fill, 10% O
2
, 6 rpm, 100
NL/min.
5.4.5 Effect of raw material bed fill degree
The effect of the raw material bed fill degree on the conversion time has been investigated for tire
rubber, wood, petcoke and polypropylene. Figure 5-12 left shows the conversion time for
cylindrical tire rubber particles at volumetric fill degrees of 0% (empty rotary drum), 5% and 10%
at 900C and 10 vol.-% oxygen. The overall carbon conversion times are observed to be highly
influenced by the fill degree: In an empty rotary drum, the conversion time is around 100 s while it
is around 200 s at 5% fill degree. For the highest fill degree of 10%, the conversion time is 600 s,
significantly longer than for the other two cases. It is also interesting to note that the shape of the
curves are quite different, indicating that the fraction of volatile carbon and char carbon are
different depending on the drum fill degree. The reason may be differences in the heating
mechanism, depending on whether the tire cylinder is buried in the bed or only exposed to the wall
and freeboard gas. It has previously been reported that the heating rate of tire rubber has a
significant effect on the volatile/char ratio (Leung and Wang, 1998).
99
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
88
0 200 400 600
Time [s]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
0% fill
5% fill
10% fill
0 5 10 15 20 25
Time [min]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
0% fill
5% fill
10% fill
Figure 5-12: Carbon conversion as a function of raw material fill degree. Left: Tire rubber particle with dimensions: D
= 9 mm, L = 25 mm. Right: Pine wood particle with dimensions: 60x30x10 mm. Coarse sand. T = 900C, 10% O
2
, 6
rpm, 100 NL/min.
Figure 5-12 right shows the conversion time for a pine wood particle at fill degrees of 0% (empty
rotary drum), 5% and 10%. The devolatilization time is observed to be identical for the three cases
while the char oxidation time increases with increasing fill degree. Thus the raw material fill degree
is of importance for the char oxidation time, which is likely to be due to diffusion limitations of
oxygen to the char particle when fully or partly covered with raw materials.
Similar experiments conducted with 2-4 mm petcoke particles, containing almost exclusively char,
showed a tendency for increasing conversion time when the bed fill degree increased from 0% to
5%. However, it was observed that the petcoke conversion time only changed slightly when the fill
degree increased from 5% to 10%. This observation was seen regardless of the oxygen
concentration. Experiments with polypropylene, containing exclusively volatiles, showed no or
little difference in conversion time when the bed fill degree increased. Thus the overall tendencies
regarding the raw material fill degree indicate that the effect on conversion time strongly depends
on the fuel type. The reason why devolatilization rate is independent of fill degree may be due to
the fact that the fuel is swiftly buried in raw material, see chapter 4, and that the heat transfer rate
therefore is independent of fill degree.
100
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
89
5.4.6 Effect of oxygen concentration
The effect of the oxygen concentration on the conversion time for the fuels tire rubber, wood and
petcoke has been investigated in the interval 5 vol.-% to 21 vol.-%.
Figure 5-13 left shows the conversion time for cylindrical tire rubber particles as a function of the
oxygen concentration. It is seen that the conversion time increases when the oxygen concentration
decreases. The increase is approximately linear with conversion times of 140, 170, 200 and 380 s at
21, 15, 10 and 5 vol.-% O
2
.
0 100 200 300 400
Time [s]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
21 vol.-% O2
15 vol.-% O2
10 vol.-% O2
5 vol.-% O2
0 10 20 30 40
Time [min]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
21 vol.-% O2
15 vol.-% O2
10 vol.-% O2
5 vol.-% O2
Figure 5-13: Carbon conversion as a function of oxygen concentration. Left: Tire rubber particle with dimensions: D =
9 mm, L = 25 mm. Right: Pine wood particle with dimensions: 60x30x10 mm. Coarse sand, 5% fill. T = 900C, 6 rpm,
100 NL/min.
Figure 5-13 right shows the conversion time for pine wood particles as a function of the oxygen
concentration. It is observed that the oxygen concentration has practically no effect on the
devolatilization time, but a significant effect on the char oxidation time. Full conversion is reached
in less than 10 minutes at the highest oxygen concentrations of 21% and 15% O
2
, respectively. At
10% O
2
the conversion time is 18 minutes and at 5% O
2
is the conversion time 40 minutes. Thus the
oxygen concentration clearly has a significant effect on the conversion time of pine wood char, and
the conversion time increases approximately linear as the oxygen concentration decreases.
The effect of oxygen concentration was also studied for petcoke, where a linear relationship
between conversion time and oxygen concentration was also found, as is the case with pine wood
and tire char.
Overall, it can be concluded that the oxygen concentration is an important parameter for the
conversion time of the char because the char oxidation is dependent on oxygen diffusion to the char.
101
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
90
5.4.7 Effect of rotational speed
The effect of drum rotational speed on the conversion time has been studied for tire rubber, wood
and petcoke. Figure 5-14 left shows the conversion time for cylindrical tire rubber particles as a
function of rotational speed in a bed with 5% fill degree at 900C and 10 vol.-% O
2
. It is observed
that the conversion time increases when the rotational speed decreases. The conversion time is e.g.
100 s at 20 rpm and 200 s at 6 rpm, a factor 2 in difference. The conversion time increases
dramatically to 440 s at the lowest rotational speed of 3 rpm. The reason is likely to be a change in
bed behavior, which could also be observed visually: The bed motion is slumping at 3 rpm while it
is rolling at 6 rpm and higher rotational speeds. The difference between slumping and rolling bed
motions was illustrated in Figure 4-1. This difference in bed motion affects the mixing process and
thus the conversion time of the fuel particles. It is also possible that the fuel particle break-down is
more significant at higher rotational speeds due to stronger forces acting on the fuel particles.
0 100 200 300 400 500
Time [s]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
20 rpm
15 rpm
10 rpm
6 rpm
3 rpm
0 10 20 30 40
Time [s]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
20 rpm
15 rpm
10 rpm
6 rpm
3 rpm
Figure 5-14: Carbon conversion as a function of rotational speed. Left: Tire rubber particle with dimensions: D = 9
mm, L = 25 mm. Right: Pine wood particle with dimensions: 60x30x10 mm. Coarse sand, 5% fill. T = 900C, 10% O
2
,
100 NL/min.
Figure 5-14 right shows the conversion time for pine wood particles as a function of rotational
speed in a bed with 5% fill degree at 900C and 10 vol.-% O
2
. It is observed that there is no
difference for the devolatilization time. For the char oxidation, the conversion time increases
slightly when the rotational speed is decreased in the interval 20 rpm to 6 rpm. But when the
rotational speed is lowered to 3 rpm, then the conversion time is increased significantly (33 minutes
relative to 17 minutes at 6 rpm). This large effect at the lowest rotational speed is due to a change in
bed behavior, from rolling to slumping.
Experiments with petcoke in the interval 3 rpm to 20 rpm also showed that the conversion time
increases as the rotational speed decreases. This tendency is observed for all tested rotational speeds
with differences in conversion times of up to 2 minutes between one rotational speed and the next
tested rotational speed.
Overall the results indicate that the rotational speed has an effect on the conversion time,
particularly if there is a shift in the dominant bed motion when going from one to another rotational
speed.
102
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
91
5.4.8 Effect of temperature
The effect of temperature on the conversion time has been investigated for the fuels tire rubber and
pine wood for the temperatures 700C, 800C, 900C and 1,000C.
Figure 5-15 left shows the conversion time of tire rubber as a function of temperature. It is observed
that both the devolatilization time and the char oxidation time decreases when the temperature
increases. The conversion time at 1,000C is observed to be around 160 s, while it is 490 s at
700C.
0 100 200 300 400 500
Time [s]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
1000C
900C
800C
700C
0 10 20 30 40
Time [min]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
1000C
900C
800C
700C
Figure 5-15: Carbon conversion as a function of temperature. Left: Tire particle with dimensions: D = 9 mm, L = 25
mm. Right: Pine wood particle with dimensions: 60x30x10 mm. Coarse sand, 5% fill. 6 rpm, 10% O
2
, 100 NL/min.
Figure 5-15 right shows the conversion time for pine wood particles as a function of temperature in
the temperature interval 700C to 1,000C. Both the devolatilization time and the char oxidation
time increases when the temperature decreases. However, the time difference for the char oxidation
at 700C and 800C is quite small, 0 to 1 minute, whereas the time is significantly shorter at 900C
and 1,000C.
The effect of temperature on the conversion time of polypropylene flakes has also been studied. The
result was that the conversion time decreased as the temperature increased, similarly as for tire
rubber and pine wood.
103
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
92
5.4.9 Conclusions for experimental parameter study
A parameter study has been conducted for four selected fuels; tire rubber, polypropylene, petcoke
and pine wood.
Fuel particle size and shape were investigated for tire rubber and pine wood. The result was that the
fuel particle thickness and external surface area is important for the conversion time, both in terms
of devolatilization and char oxidation.
For polypropylene (and the devolatilization part of the other fuels) temperature is the rate
determining parameter.
For the char oxidation part of tire rubber, wood and petcoke, the parameters char mass, oxygen
concentration, temperature, raw material fill degree and raw material characteristics are all
important parameters, but the rotational speed was also observed to have an effect on the conversion
time.
104
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
93
5.5 Model analyses
This section seeks to develop and verify mathematical models for devolatilization and char
oxidation of selected alternative fuels. The models are to be used under conditions similar to those
in the material inlet end of cement rotary kilns and may be modified to predict fuel conversion
times in cement rotary kilns. This industrial modelling will be described in greater detail in chapter
7.
5.5.1 Model for devolatilization
The heat up of a large, spherical particle may be derived by solving the unsteady heat transfer
differential equation:
2
2
1
p p p
p p
T k T
r
t r r C r
| | c c
c
=
|
|
c c c
\ .
(5.3)
with the following boundary and initial conditions:
( ) ( ) ( ) ( ) ( ) ( )
4 4 4 4
1 for r = R
p
p Cond b p Conv g p p g g g p p w w w p
T
k Yh T T Y h T T T T T T
r
oc c o oc c o
c
= + + +
c
(5.4)
0 for r = 0
p
T
r
c
=
c
(5.5)
where R is the initial radius of the particle, k
p
is the thermal conductivity of the particle,
p
is the
virgin fuel density and C
p
is the specific solid heat capacity. h
Cond
and h
conv
are the heat transfer
coefficients for conduction and convection, respectively. T
p
, T
b
, T
g
and T
w
are temperatures of
particle, bed, bulk gas and inner kiln walls, respectively. Y is the dimensionless distribution of fuels
in the bed, as illustrated in Figure 5-2. Linjewile et al. (1993) studied the heat transfer to large 6-10
mm diameter petcoke particles from a fluidized bed of sand of particle sizes _ 1 mm. The
conductive heat transfer coefficients were in the range of 200-500 W/m
2
K. Convective heat transfer
under different kiln operating parameters in pilot plant scale has been studied by Tscheng and
Watkinson (1979). Based on their experimental results with sand and limestone, correlations have
been proposed for convective heat transfer coefficients, h
Conv
.
Equation (5.4) expresses that all heat transferred to the surface of the particle is conducted into the
particle, and equation (5.5) expresses that the gradient at the particle center is zero due to symmetry.
In equation (5.4) it should be realized that the existence of a flame front around the fuel particle
may increase T
g
and T
p
locally. Wildegger-Gaissmaier et al. (1990) have suggested a correlation for
the temperature increase due to a flame front during volatile combustion around large coal particles.
However, when implementing this correlation in the devolatilization model in the present study,
significantly faster devolatilization times were found, which corresponded poorly with the
experimentally found devolatilization times. The correlation for the flame front is therefore not
included in equation (5.4).
Equation (5.3) must be solved numerically. However, an analytical solution has been suggested by
Agarwal et al. (1986):
105
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
94
( )
2 2
,0
/ ,0
1
,0
,0
sin
sin cos
2
sin cos
i p
p
i
t r p a p
i i i
i
a p i i i p
i
p
r
r T T
e
T T r
r
| o
|
| | |
| | |
|


=
( (
( (

( (

=
(
(
(
(
(
(

(5.6)
where T
a
is the mean average temperature surrounding the particle. T
a
may be estimated as T
a
= Y
T
b
+ (1-Y) T
g
. is the thermal diIIusivity, k
p
/(
p
C
p
).
The
i
s are the positive solutions to the equation:
cos (1-Bi) sin (5.7)
where Bi is the Biot number. An efficient heat transfer coefficient, h
eff
, is used to calculate Bi,
where h
eff
= Y h
Cond
+ (1-Y) (h
Conv
+ h
rad
), and h
rad

g
(T
g
2
+ T
p
2
) (T
g
+ T
p
) (Szekely et al.,
1976).
The time,
devol
, required for complete devolatilization corresponds to the time needed for the fuel
particle centre to reach the upper devolatilization temperature, T
Vol2
, as described in section 5.1.3.
This time is determined from equation (5.6) by using T
p
(0, t) = T
Vol2
(Agarwal et al., 1984).
However, equation (5.6) should only be used to determine the devolatilization time for an initially
dry particle, since the equation does not take the particle moisture content into consideration. The
moisture content of a fuel particle may influence the devolatilization time considerably (Agarwal et
al., 1984 and 1986, and de Diego et al., 2002). Agarwal et al. (1984) studied the drying of wet coals
during fluidized bed combustion and proposed a correlation for the total drying time:
( )
1.09
1.1
2
0
0.14 4
Drying p
eff P b e
Bi
C r
Bi C T T

t
o
(
| |
+ | |
(
=
|
|
|
( \ .
\ .

(5.8)
where
eff
is the effective thermal diffusitivity, C
0
is the moisture content, C
p
is the fuel heat
capacity, is the latent heat oI vaporization, r
p
is the particle radius, T
b
and T
e
are the bed
temperature and temperature at the wet/dry interface, respectively. This correlation fitted the drying
time within 15% for 0.3 _ C
0
_ 1.8 g water/g dry Iuel, 1 _ Bi _ 20, 400 _ T
b
_ 1400 K.
If drying and devolatilization is considered to be sequential processes, then equations (5.8) and (5.6)
may be used to calculate the drying time and the devolatilization time. This approach is suitable for
low Biot numbers because the tendency of the particle to be isothermal is greater. But for higher
Biot numbers, the devolatilization may start before the drying is completed, and a combined
approach is thus required (Agarwal et al., 1986). Several analytical and numerical procedures have
been suggested in the literature, as summarized by Agarwal et al. (1986) and de Diego et al. (2002).
de Diego et al. (2002) studied devolatilization of wet pine wood particles and found a useful
correlation, based on a modified power-law relationship, which included both moisture content,
temperature and particle size:
106
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
95
( ) ( ) ( )
2
1.6 0.1
0
1 1.7
devol sph H O
a T d y t u

= + (5.9)
where d
sph
is the equivalent spherical diameter oI the Iuel particle in mm and is the sphericity of
the fuel particle, see section 5.5.2 for definitions of d
sph
and . y
H2O
is the weight fraction of water in
the fuel particle. a
0
is a temperature dependent constant which was found to decrease approximately
linearly with temperature: a
0
= 1.89, 1.38, 1.03 and 0.89 at 650C, 750C, 850C and 950C,
respectively. This correlation gave excellent fits with experimentally found devolatilization times
for cubic and rectangular pine wood particles of thicknesses from 4 to 15 mm. The moisture content
was up to 50 wt.-%.
5.5.2 Validation of devolatilization model
In section 5.5.1 an analytical expression for the heating of large fuel particles was given, see
equation (5.6). If the temperature required for complete devolatilization, T
Vol2
, is known, e.g. from
TGA-measurements reported in literature, the analytical expression may be used to determine the
time required for full devolatilization. Table 5-4 shows the temperatures for initial and final
devolatilization temperatures for the fuels tire rubber and pine wood.
Fuel T
Vol1
[K] T
Vol2
[K] Reference
Tire rubber 523 773 Chinyama et al. (2007)
Pine wood 473 773 Kim et al. (2010)
Table 5-4: Temperatures for initial and final devolatilization of tire rubber and pine wood.
Experiments with non-spherical tire rubber and pine wood particles of different size and shapes
have been conducted to study the devolatilization time. Since the model assumes that the fuel
particles are spherical, it is necessary to represent the non-spherical fuel particles as spherical
particles of diameter, d
eff
, characterized by the non-spherical and spherical particles having the same
total surface area (Kunii and Levenspiel, 1991):
eff sph
d d u = (5.10)
where d
sph
is the equivalent spherical diameter, defined as the diameter of a sphere with the same
volume as the particle. is the sphericity deIined as:
of same volume
Surface of sphere
Surface of particle
u
| |
=
|
\ .
(5.11)
The experimentally found devolatilization times are determined by means of the normalized carbon
conversion curves for the fuels. Figure 5-16 shows an example with pine wood, where the
devolatilization time is observed to begin at time 0 s and end at time 80 s. The devolatilization ends
when the slope of the conversion curve is observed to decrease significantly, indicating the start of
the slower char oxidation. It is relatively straight-forward to determine devolatilization times for
pine wood but more complex for tire rubber: Tire rubber shows no clear end point for
devolatilization, see e.g. Figure 5-10 left. For tire rubber the end of devolatilization has been
defined when the normalized carbon conversion is 0.7.
107
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
96
Figure 5-16: Determination of devolatilization time for a pine wood particle at 900C and 10 vol.-% O
2
. 5% volumetric
fill, coarse sand, 100 NL/min, 6 rpm. Particle dimensions: 60 x 30 x 10 mm (7.0 g).
Devolatilization of tire rubber
A comparison between experimental values and values predicted by the model for tire rubber is
shown in Figure 5-17. Values used in the model can be found in Appendix B. In the model the
fraction of the fuel particle surface that is covered by raw materials, Y, is 0.2. This fraction was
chosen based on visual observations during the experiments, where the cylindrical tire rubber
particle was observed to be free of the raw material bed during most of the devolatilization. The
experimental values were obtained for four different tire rubber particle diameters from 10 mm to
18.5 mm and at 900C. The two smallest tire rubber particles were also studied at 700C, 800C
and 1,000C. The model is in acceptable agreement with the experimental values, which were found
to have devolatilization times from approximately 40 to 120 seconds, depending on temperature and
tire rubber particle size. The model generally overestimates the effect of temperature compared to
the experimental findings: For the two smallest fuel particles, the difference in experimentally
found devolatilization times was smaller than predicted by the model.
0 200 400 600 800 1000
Time [s]
0
0.2
0.4
0.6
0.8
1
N
o
r
m
a
l
i
z
e
d
c
a
r
b
o
n
c
o
n
v
e
r
s
i
o
n
Devolatilization start
Devolatilization end
108
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
97
Figure 5-17: Comparison of tire rubber devolatilization time predicted by model and by experimentally found values.
10% O
2
. 5 volumetric Iill degree in rotary drum. Y 0.2. 1.25 10
-7
m
2
/s.
Chinyama and Lockwood (2007) studied the devolatilization times of tire rubber particles with
thicknesses in the range 6-12 mm and in the temperature interval from 700-1,000C. The
devolatilization times were found to be 30-100 seconds depending on thickness and temperature,
which were typically 20 seconds longer than the values found in this study for the same particle
sizes and temperatures. This difference in times may be due to different heat transfer mechanisms at
the two experimental conditions: In the experiments conducted by Chinyama and Lockwood
(2007), heat transfer to the tire rubber particles were exclusively by convection and radiation,
whereas heat transfer in this study were also partly by conduction from the raw materials and drum
wall. Another reason may be the mechanical influence from the rotary drum in this study, which is
likely to increase fragmentation of the fuel particle during devolatilization in fact, the tire rubber
particles were observed to fragment a little during the experiments, thereby leading to smaller fuel
particle sizes.
Larsen et al. (2006) also studied the devolatilization of cylindrical tire rubber particles with
diameters from 7 to 22 mm and heights of 35 mm at temperatures up to 840C in an inert
atmosphere. The reported devolatilization times at 840C were 75 to 300 s when increasing the
particle diameter from 7.5 to 22 mm. These devolatilization times are approximately twice as long
as indicated by the results from this study. The reason for the deviations may again be differences in
the heat transfer mechanisms due to different experimental conditions as well as the mechanical
influence in this study. In addition, the experiments conducted by Larsen et al. (2006) were made
under an inert atmosphere while the present study was made under oxidizing conditions which are
likely to have accelerated the devolatilization rate.
109
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
98
Devolatilization of pine wood
A comparison between experimentally found values and values predicted by the model for pine
wood is shown in Figure 5-18. Values used in the model can be found in Appendix B. The
experimental values were obtained for five different pine wood particle diameters from 12 mm to
26 mm and at 900C. The pine wood particles of diameters 15 and 20 mm were also studied at
700C, 800C and 1,000C. The model is observed to be in acceptable agreement with the
experimental values, which were found to have devolatilization times from approximately 40 to 170
seconds, depending on temperature and particle size. However, the experiments conducted at 700C
and 800C are generally 20-30 seconds longer than the devolatilization times predicted by the
model and in one case, with a particle diameter of 20 mm and at 700C is the deviation 50 seconds.
Figure 5-18: Comparison of pine wood devolatilization time predicted by model and by experimentally found values.
10% O
2
, 5 volumetric Iill degree in drum. Y 0.2. 1.9 10
-7
m
2
/s.
Devolatilization of non-spherical pine wood particles with equivalent diameters, d
eff
, in the range 10
to 45 mm and in the temperature interval 650 to 850C has been investigated by de Diego et al.
(2002). The reported devolatilization times were in good correspondence with the values found in
the present study. For example, at 850C the devolatilization times for 10 to 30 mm diameter
particle were reported to be from 30 s to 150 s, whereas the results from this study indicate
devolatilization times from 25 s to 180 s at 850C. These minor differences in devolatilization times
may be due to different characteristics of the pine wood used in the experiments. The correlation for
pine wood devolatilization suggested by de Diego et al. (2002) and described in section 5.5.1,
equation (5.9), also fits the experimental results from this study nicely, and may be used as a
simplified model to predict pine wood devolatilization times.
110
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
99
The found devolatilization times for pine wood are also in relatively good consistency with
devolatilization times for beech wood particles of similar size as reported by Di Blasi and Branca
(2003) and Jand and Foscolo (2005), see the literature study in section 5.2.2.
5.5.3 Model for char combustion
The aims of this section are to derive a model for char conversion in the rotary drum, including the
main rate limiting parameters for char conversion. In section 5.5.4 the char combustion model will
be validated against experimental results, and in chapter 7 the model will be modified into a
realistic char conversion model for the material inlet end of industrial rotary kilns. The char
conversion model may be used in combination with models for drying and devolatilization of virgin
fuel particles, or it may stand alone to simulate the fate of char particles dropping from calciners
through the kiln riser duct and into the material inlet end of rotary kilns.
Figure 5-19 shows a rotary kiln where char particles are buried in the raw material bed. The bulk
oxygen volume fraction is denoted y
O2,
, and the oxygen volume fraction at the bed surface is
called y
O2,surf
. The oxygen volume fraction at the char particle surface is called y
O2,part
. The mixing
of large particles into a bed of smaller particles was shown in chapter 4 to be a fast process,
typically taking only a few bed revolutions.
Figure 5-19: Oxygen mass transfer to char particles in a bed of cement raw materials in a rotary kiln.
In a cement rotary kiln it may be assumed that mass transfer of oxygen into the active layer is fast
while oxygen mass transfer into the passive layer takes place by a relatively slow diffusion, see
Figure 5-19. Since the char particles will continuously be transported through the active and passive
layers, respectively, it may be assumed that char oxidation primarily takes place when the char
particles are present in the active layer.
With the above mentioned considerations in mind, a model for char oxidation applicable for the
conditions in the material inlet end of a rotary kiln should include the following:
111
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
100
1. External mass transfer of oxygen to bed surface.
2. Mass transfer of oxygen through the active layer.
3. Diffusion of oxygen into the char particle.
4. Chemical reaction of char and oxygen.
The external mass transfer of oxygen to the bed surface can be written as the mass transfer
coefficient k
g
multiplied by the driving force:
( )
2 2
2
, , 2
mol O
m s
g O O surf
g
P
r k y y
RT

(
=
(


(5.12)
where k
g
is found from a Sherwood correlation. For the rotary drum, where the gas flow is in the
laminar regime, the following correlation is used (Green and Perry, 2008):
2
1/2 1/3
0.646Re
g
O
k L
Sh Sc
D
= = (5.13)
where Re is calculated from the rotary drum length, L. This correlation is valid for laminar flow
over a flat plate, which is roughly similar to the situation with gas flow above the bed surface.
For an industrial rotary kiln where the gas flow is in the turbulent regime, this correlation is used
(Green and Perry, 2008):
2
0.83 0.44
0.023Re
g i
O
k D
Sh Sc
D
= = (5.14)
Re is calculated from the rotary kiln diameter, D
i
. This correlation is valid for turbulent flow in
pipes.
In case of a rotary kiln/drum without raw materials and few fuel particles, the mass transfer
coefficient may be found from a Sherwood correlation for flow past single spheres (Green and
Perry, 2008):
2
1/2 1/3
2 0.6 Re 2 Re 800, 0.6 Sc 2.5
p p
g p
d d
O
k d
Sh Sc
D

= = + s s s s (5.15)
The mass transfer of oxygen from the bed surface into the active layer can be written as the mass
transfer coefficient in the active layer, k
AL
, multiplied by the driving force and the probability that
the char particles are in the active layer, P
AL
, at any given time:
( )
2 2
2
, ,part surf 2
mol O
m s
AL AL O surf O
b
P
r P k y y
RT
(
=
(


(5.16)
112
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
101
The probability for the char particles to be in the active layer may be estimated by the
experimentally determined correlation suggestion by equation (4.5) in chapter 4, which gave a
correlation for the percentage of visible fuel particles at the bed surface. This correlation may be
used since the active layer thickness is quite small and therefore it may be assumed that the number
of fuel particles in the active layer is practically the same as the number of fuel particles that are
visible above the bed surface. The mass transfer coefficient in the active layer, k
AL
, has been
determined by Heydenrych (2001):
( )
2
2
, 2
6 1
, 3.8
O
AL O eff
p
Sh D
k D Sh
d
c
= = (5.17)
where D
O2, eff
is the effective diffusion coefficient of oxygen:
2 2
, O eff O
D D
c
t
= (5.18)
D
O2
is the diIIusion coeIIicient Ior oxygen in air, is the bed tortuosity and is the bed porosity or
void Iraction available Ior gas Ilow. Dias et al. (2006) Iound that the tortuosity, , oI a mixed bed oI
granular particles could be described as 1/
n
, where n depends on the particle packing but is
usually in the range of 0.4-0.5.
The rate of oxygen diffusion into the char particle may be estimated by the expression:
( )
( )
2
2 2
,
2
,part surf ,part core 2
mol O
m s
O part
O O
p p
D
P
r y y
r t RT
(
=
(


(5.19)
where D
O2, part
is the intra-particle oxygen diffusion coefficient, r
p
is the particle radius, y
O2, part surf
and y
O2, part core
are the oxygen volume fractions at the particle surface and at the intra-particle
ash/char interface, respectively. D
O2, part
may be calculated from:
( )
2 2 2 2 2
0
0 0
,
1
1 1
Vol ash
part
Vol ash
O part O O O O
part part part part
y y
y y
D D D D D

c

t t t t


+
~ ~ ~ ~ (5.20)
where the char particle porosity,
part
, is calculated by subtraction of the weight fractions of volatiles
and ash, y
Vol
and y
ash
. The char particle tortuosity for chars is typically around 3 to 5 (Johnsson and
Jensen, 2000).
The rate of reaction at the unreacted char outer surface between char and oxygen is found from the
intrinsic rate expression multiplied with the total moles and divided by the char surface area and
product distribution ratio between CO and CO
2
:
( )
2
2
0 ,part core 2
mol O
exp 1
m s
m
n C a
O
char p
n E
r k y X
A RT ,
| |
(
=
|
(
|


\ .
(5.21)
113
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
102
where is the product distribution ratio between CO and CO
2
, X is the char fractional conversion
degree and A
char
is the char surface area. A correlation between temperature and the CO/CO
2
ratio,
, during combustion of different coal chars has been reported by Arthur (1951):
3.4
2
51916 /
10 exp
CO J mol
CO R T
,
| |
= =
|

\ .
(5.22)
This correlation may be used to estimate at different temperatures under the assumption that it is
representative for other chars than coal chars.
At pseudo steady state, the O
2
flux balances the consumption at the char particle surface and the
char conversion may be expressed as:
mol C
C
surf
dn
A r
dt s
,
(
=
(

(5.23)
The conversion may also be expressed in terms of fractional conversion, X, instead of moles, n
C
, by
inserting:
,0
1
C
C
n
X
n
= (5.24)
and
,0
1
C
C
dX dn
n
= (5.25)
The following expression is then obtained:
,0
surf
C
A r
dX
dt n
,
= (5.26)
The unknown oxygen concentrations y
O2,surf
, y
O2,part surf
and y
O2,part core
may be found by solving the
four equations (5.12), (5.16), (5.19) and (5.21) for the four unknowns r, y
O2,surf
, y
O2,part surf
and
y
O2,partcore
. This is easily done for n = 1 in equation (5.21), whereas a numerical solution procedure is
required for n = 1. For n = 1, the following expression for char conversion is derived:
( )
( )
2
2
,
1/3
,0
,0
1 ,
,0 0
1
1 exp
O surf
C
g p p
b char
m g AL AL O part
a
C
p
y A
dX
dt n
R T r X R T
R T A
k P P k P D P
E
n X k
R T
,
,

=


+ + +
| |

|
|

\ .
(5.27)
This expression includes the resistances external oxygen transport, oxygen transport into the bed,
diffusion into the char particle and chemical reaction between char and oxygen. The intra-particle
114
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
103
diffusion term may be omitted for small char particles, or for char particles that rapidly fragments
into many smaller char particles. It may also be shown that the term for char oxidation kinetics
becomes negligible at high temperatures above 900C. Fuel specific, kinetic data and densities used
in the model are summarized in Appendix B. Appendix C includes calculations on char density,
oxygen concentration and the diffusion coefficient.
5.5.4 Analysis and comparison of char combustion model
This section analyzes the char combustion model under the experimental conditions in the rotary
drum and compares the model predictions with experimental results. The purpose is to study effects
of mass, oxygen concentration, bed fill degree and temperature on the char conversion time.
Figure 5-20 left shows the char conversion times for three different pine wood char particles of
masses of 0.12 g, 0.24 g and 0.88 g, respectively. The initial fuel particle dimensions are 30 x 10 x 8
mm, 30 x 10 x 15 mm and 60 x 30 x 10 mm, respectively. The oxygen concentration is 5, 10 or 21
vol.-% O
2
. It is observed that the model, shown with solid lines, predicts a linear relationship
between conversion time, char mass and oxygen concentration. The experimental data also confirm
this linear relationship: The conversion time for the 0.88 g char particle is for example around 300
s, 600 s and 1300 s at oxygen concentrations of 21 vol.-%, 10 vol.-% and 5 vol.-% O
2
. As shown in
Figure 5-20, left the model is generally observed to give a good fit to the experimental data in the
experiments with different char masses and oxygen concentrations.
.
0 0.2 0.4 0.6 0.8 1
Char mass [g]
0
400
800
1200
T
o
t
a
l
c
h
a
r
c
o
n
v
e
r
s
i
o
n
t
i
m
e
[
s
]
5 vol.-% O2
10 vol.-% O2
21 vol.-% O2
Model
(21 vol.-% O
2
)
Model
(10 vol.-% O
2
)
Model
(5 vol.-% O
2
)
0 5 10
Volumetric bed fill degree [%]
0
200
400
600
800
1000
1200
T
o
t
a
l
c
h
a
r
c
o
n
v
e
r
s
i
o
n
t
i
m
e
[
s
]
700C
800C
900C
1000C
Increasing temperature, model
(700C, 800C, 900C and 1000C)
Figure 5-20: Comparison of experimental data with model predictions for pine wood. Left: Effect of char mass and
oxygen concentration on the char conversion time. 900C, 5% fill degree, 100 NL/min. Initial particle dimensions: 30 x
10 x 8 mm, 30 x 10 x15 mm and 60 x 30 x 10 mm. Right: Effect of raw material fill degree and temperature on model
predicted char conversion time for a pine wood char particle with initial dimensions of 60 x 30 x 10 mm. Char mass =
0.88 g. 10 vol.-% O
2
, 100 NL/min. Char conversion evaluated at 80% conversion degree.
Figure 5-20 right shows the char conversion time as a function of volumetric bed fill degree and
temperature. The fill degrees are 0% (only char in the reactor), 5% or 10%, and the temperatures are
700C. 800C, 900C or 1000C. It is observed that the char conversion time increases when the
115
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
104
bed fill degree increases. The model shows the same tendencies in temperature dependence as the
experimental results. However the experimental data show greater differences in conversion time as
a function of temperature than the model, particularly at 700C where the deviation is up to 350 s at
10% fill degree corresponding to a deviation of around 30%.
Figure 5-21 left shows the char conversion time for three different tire rubber char cylinders with
masses of approximately 0.6 g, 1.4 g and 2.0 g, respectively. The oxygen concentration is 5, 10 or
21 vol.-% O
2
. The model, shown with solid lines, predicts a linear increase in conversion time with
char mass and a linear decrease with oxygen concentration. For example, the predicted conversion
time for 2 g of char is observed to be halved from 570 s to 280 s when the oxygen concentration is
doubled from 5 vol.-% to 10 vol.-%. Thus the model predicts that external mass transfer of oxygen
is a main rate limiting parameter. The model gives the best agreement with the experimental values
at 10 vol.-%and 21 vol.-%O
2
, where the deviation is generally within 50 s. But the model fails to
give good agreement with the experiments at 5 vol.-% O
2
and char masses of 1.4 g and 2.0 g. In
these cases, the deviations are up to nearly 30%.
0 0.5 1 1.5 2 2.5
Char mass [g]
0
100
200
300
400
500
600
T
o
t
a
l
c
h
a
r
c
o
n
v
e
r
s
i
o
n
t
i
m
e
[
s
]
5 vol.-% O2
10 vol.-% O2
21 vol.-% O2
Model
(21 vol.-% O
2
)
Model
(10 vol.-% O
2
)
Model
(5 vol.-% O
2
)
0 5 10
Volumetric bed fill degree [%]
0
100
200
300
400
500
T
o
t
a
l
c
h
a
r
c
o
n
v
e
r
s
i
o
n
t
i
m
e
[
s
]
700C
800C
900C
1000C
Increasing temperature, model
(700C, 800C, 900C and 1000C)
Figure 5-21: Left: Comparison between model predictions and experimental data for char conversion time of three
different tire rubber char cylinders with lengths of 25 mm and initial diameters of 9, 12 and 16 mm, respectively.
900C, 5% raw material fill, 100 NL/min. Char masses and oxygen concentrations indicated in figure. Right: Effect of
raw material fill degree and temperature on model predicted char conversion time for tire char cylinder with a length of
25 mm and initial diameter of 9 mm. Char mass = 0.6 g. 10 vol.-% O
2
, 100 NL/min. Char conversion evaluated at 80%
conversion degree.
Figure 5-21 right shows the predicted effect of raw material fill degree in the interval 0-10% and in
the temperature interval 700C to 1000C for a cylindrical tire rubber char particle. Experimental
values are also compared with the model. Both experimental values and the model indicate that the
char conversion time increases with volumetric bed fill degree and decreases when the temperature
increases. The model only fits the experimental values in the case with a fill degree of 0%,
corresponding to a situation with only char in the rotary drum and no cement raw materials.
However, the experimental values at 700C and 0% fill show longer conversion times than
predicted by the model: 120-130 s relative to 65 s, respectively. At 5% fill degree, the model only
116
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
105
fits the experimental values at 900C and 1000C, but under-estimates the conversion times at
700C and 800C by a factor of 2. And at a fill degree of 10%, the model under-estimate the effect
of fill degree and temperature by a factor of 2.5 to 3, depending on temperature. The effect of fill
degree seems to increase more with fill degree for tire rubber than for pine wood. The reason for
these differences may be related to the chemical and physical properties of the different chars: Tire
char fragments into several small particles while the wood char to a high degree retains its size and
shape.
Conclusions for char conversion model
A model for char conversion has been suggested and compared with experimental data for tire char
and pine wood char. The model includes the four resistances external oxygen diffusion to raw
material bed, oxygen transport into the bed, intra-particle oxygen diffusion and chemical reaction at
the char surface. By analysis of the char conversion model, equation (5.27), it was found that most
resistance was due to external oxygen diffusion and oxygen transport into the bed. Intra-particle
oxygen diffusion and chemical reaction kinetics were found to be minor resistances to char
oxidation.
The model is in good correspondence with the observed conversion times for pine wood char.
However, it under-estimates the effect of bed fill degree for tire char. The model also under-
estimates the effect of the lowest temperatures of 700C and 800C on the conversion time for both
pine wood char and tire char. Regarding the effect of material fill degree on the char conversion
time, it is necessary to carry out further experiments with different fuel chars and particle
sizes/shapes to obtain a better basis for model validation. Regarding the effect of temperature, it is
possible that the model does not give a realistic prediction for the reaction kinetics. It is also
possible that the set-point temperature used in the model is different from the actual surface
temperature of the char particles during char oxidation.
5.6 Conclusions for chapter 5
Combustion of a solid fuel particle consists of the following steps: heating, drying, devolatilization
and char oxidation. Ash is the ultimate product left after complete combustion. In a cement rotary
kiln where gas and solid temperatures are typically higher than 900C, heating, drying and
devolatilization are typically rapid processes, whereas char oxidation may be a significantly slower
process. The reason is that char oxidation is dependent on oxygen transport to the char particle
before the oxidation can take place.
Experiments with different solid fuels in a high-temperature rotary drum have been conducted in
order to quantify the effect of key process parameters such as bed fill degree, drum rotational speed,
temperature, etc. on the fuel conversion time. The process conditions have been chosen to simulate
the process conditions in the material inlet end of an ILC rotary kiln. The main focus has been on
tire rubber and pine wood.
The results showed that devolatilization of tire rubber and pine wood was mainly influenced by the
temperature and fuel particle size. The char oxidation was influenced by all investigated parameters,
i.e. fuel sample mass, particle size, temperature, oxygen concentration, bed fill degree and rotational
speed. Rotational speed, however, was only of importance if there was a shift in bed motion, e.g.
from slumping to rolling. For changes in rotational speed where the dominant bed motion was
maintained as the rolling motion, the effect of rotational speed was of minor importance.
117
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
106
The tire char was observed to fragment into smaller particles during the char oxidation, leading to
shorter conversion times than for pine wood char which predominantly behaved as one shrinking
particle during the char oxidation.
Devolatilization and char oxidation models have been developed and compared with experimental
results and published data. The devolatilization models fits well to the experimental data for pine
wood and tire rubber. The char oxidation model for pine wood char also fits well with experimental
data. However, there is a tendency that the effect of temperature is under-estimated by the char
model for pine wood char. For tire char, the char oxidation model does not give good predictions,
particularly regarding temperature and bed fill degree.
5.7 References
Agarwal, P. K., Genetti, W. E., Lee, Y. Y. and Prasad, S. N.; Model for drying during fluidized-bed
combustion of wet low-rank coals, Fuel, 63, 1020-1027, 1984.
Agarwal, P. K., Genetti, W. E. and Lee, Y. Y.; Devolatilization of large coal particles in fluidized
beds, Fuel, 63, 1748-1752, 1984.
Agarwal, P. K., Genetti, W. E. and Lee, Y. Y.; Coupled drying and devolatilization of wet coal in
fluidized beds. Chem. Eng. Sci., 41 (9), 2373-2383, 1986.
Arthur, J. R.; Reactions between carbon and oxygen. Transactions of the Faraday Society, Vol. 47,
Is. 2, 164-178, 1951.
Atal, A. and Levendis, Y. A.; Comparison of the combustion behavior of pulverized waste tyres and
coal. Fuel, Vol. 74, No. 11, 1570-1581, 1995.
Bouvier, J. M., Charbel, F. and Gelus, M.; Gas-solid pyrolysis of tire wastes kinetics and material
balances of batch pyrolysis of used tires. Resources and Conservation, 15, 205-214, 1987.
Chinyama, M. P. M. and Lockwood, F. C.; Devolatilisation behaviour of shredded tyre chips in
combusting environment. J. Energy Institute, 80, 3, 162-167, 2007.
Conesa, J. A., Font, R., Fullana, A. and Caballero, J. A.; Kinetic model for the combustion of tyre
wastes. Fuel, 77, 13, 1469-1475, 1998.
Dias, R., Teixeira, J. A., Mota, M. and Yelshin, A.; Tortuosity variation in a low density binary
particulate bed. Separation and Purification Technology, 51, 180-184, 2006.
Di Blasi, C.; Modelling the fast pyrolysis of cellulosic particles in fluid-bed reactors. Chem. Eng.
Sci., 55, 5999-6013, 2000.
Di Blasi, C. and Branca, C.; Temperatures of wood particles in a hot sand bed fluidized by nitrogen.
Energy & Fuels, 17, 247-254, 2003.
118
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
107
De Diego, L. F., Garca-Labiano, F., Gayn, P, Abad, A. and Adnez, J.; Coupled drying and
devolatilisation of non-spherical wet pine wood particles in fluidized beds. J. Analytical and
Applied Pyrolysis, 65, 173-184, 2002.
De Diego, L. F., Garca-Labiano, F., Gayn, P, Abad, A. and Adnez, J.; Modeling of the
devolatilization of nonspherical wet pine wood particles in fluidized beds. Ind. Eng. Chem. Res.,
41, 3642-3650, 2002.
Giddings, D., Pickering, S. J., Simmons, K. and Eastwick, C. N.; Combustion and aerodynamic
behavior of car tyre chips in a cement works precalciner. J. of the Inst. of Energy, September, 75,
91-99, 2002.
Green, D. W. and Perry, R. H.; Perrys Chemical Engineers Handbook, p. 5-63 to p. 5-70,
McGraw-Hill, 8
th
ed., 2008. ISBN: 978-0-07-142294-9.
Heydenrych, M. D.; Modelling of rotary kilns. PhD Thesis, University of Twente, The Netherlands,
2001. ISBN: 90-36515440.
Hupa, M.; Lecture notes from Chemistry in Combustion Processes, bo Akademi, Finland, 2008.
Jand, N. and Foscolo, P. U.; Decomposition of wood particles in fludized beds. Ind. Eng. Chem.
Res., 44, 5079-5089, 2005.
Janse, A. M. C., de Jonge, H. G., Prins, W. and van Swaaij, W. P. M.; Combustion kinetics of char
obtained by flash pyrolysis of pine wood. Ind. Eng. Chem. Res., 37, 3909-3918, 1998.
Johnsson, J. E. and Jensen, A.; Effective diffusion coefficients in coal chars. Proceedings of the
Combustion Institute, Vol. 28, 2353-2359, 2000.
Kim, S.-S., Kim, J., Park, Y.-H. and Park, Y.-K.; Pyrolysis kinetics and decomposition
characteristics of pine trees. Bioresource Technology, 101, 9797-9802, 2010.
Kunii, D. and Levenspiel, O.; Fluidization Engineering. Butterworths, Boston, MA, USA. 2
nd
ed.,
1991. ISBN: 0-409-90233-0.
Kyari, M., Cuncliffe, A. and Williams, P. T.; Characterization of oils, gases, and char in relation to
the pyrolysis of different brands of scrap automobile tires. Energy & Fuels, 19, 1165-1173, 2005.
Larsen, M. B., Schultz, L., Glarborg, P., Skaarup-Jensen, L., Dam-Johansen, K., Frandsen, F. and
Henriksen, U.; Devolatilization characteristics of large particles of tyre rubber under combustion
conditions. Fuel, vol. 85, 1335-1345, 2006.
Larsen, M. B.; Alternative Fuels in Cement Production. PhD Thesis, Technical University of
Denmark, Department of Chemical Engineering, 2007. ISBN 978-87-91435-49-8.
Larsen, M. B., Hansen, M. L., Glarborg, P., Skaarup-Jensen, L., Dam-Johansen, K. and Frandsen,
F.; Kinetics of tyre char oxidation under combustion conditions. Fuel, vol. 86, 2343-2350, 2007.
119
Chapter 5 Heat-up, devolatilization and combustion of large solid fuel particles
108
Laurendeau, N. M.; Heterogeneous kinetics of coal char gasification and combustion. Prog. Energy
Combust. Sci., Vol. 4, 221-270, 1978.
Leung, D. Y. C. and Wang, C. L.; Kinetic study of scrap tyre pyrolysis and combustion. J. Anal. &
Appl. Pyrolysis, 45, 153-169, 1998.
Levenspiel, O.; Chemical reaction engineering. John Wiley & Sons, Inc. 3
rd
edition, 1999.
Linjewile, T. M., Hull, A. S. and Agarwal, P. K.; Heat transfer to a large mobile particle in gas-
fluidized beds of smaller particles. Chem. Eng. Sci., 48, 21, 3671-3675, 1993.
Masi, S., Salatino, S. and Senneca, O.; Combustion rates of chars from high-volatile fuels for FBC
application. Fluidised Bed Combustion, Vol. 1, 135-143, 1997.
Mermoud, F., Golfier, F., Salvador, S., Van de Steene, L. and Dirion, J. L.; Experimental and
numerical study of steam gasification of a single charcoal particle. Combustion and Flame, 145, 59-
79, 2006.
Perry Chemical Engineers Handbook, 7
th
edition, Mc-Graw Hill, 1999.
Schmidthals, H.; Luftvergasung von Altreifen zur integrierten stofflichen und energetischen
Nutzung im Klinkerbrennprozess. Doktor-Ingenieur Dissertation, Fakultt fr Maschinenbau, Ruhr-
Universitt Bochum, Germany, 2001.
Shen, D. K., Gu. S., Luo, K. H., Bridgwater, A. V. and Fang, M. X.; Kinetic study on thermal
decomposition of woods in oxidative environment. Fuel, 88, 1024-1030, 2009.
Senneca, O.; Kinetics of pyrolysis, combustion and gasification of three biomass fuels. Fuel
Processing Technology, 88, 87-97, 2007.
Szekely, J., Ewans, J. W. and Sohn, H. Y.; Gas-solid reactions. New York, Academic Press, 51,
1976.
Tscheng, S. H. and Watkinson, A. P.; Convective heat transfer in a rotary kiln. Can. J. Chem. Eng.,
Vol. 57, 433-443, August 1979.
Wildegger-Gaissmaier, A. E. and Agarwal, P, K.; Drying and devolatilization of large coal
particles. Fuel, 69, 44-52, 1990.
Winter, F., Prah, M. E.and Hofbauer, H.; Temperatures in a fuel particle burning in a fluidized bed:
The effect of drying, devolatilization and char combustion. Combustion and Flame, 108, 302-314,
1997.
Yang, J., Tanguy, P. A. and Roy, C.; Numerical model for the vacuum pyrolysis of scrap tires in
batch reactors. AIChE Journal, Vol. 41, No. 6, 1500-1512, June 1995.
Yang, J., Tanguy, P. A. and Roy, C.; Heat transfer, mass transfer and kinetics study of the vacuum
pyrolysis of a large used tire particle. Chem. Eng. Sci., Vol. 50, No. 12, 1909-1922, 1995.
120
Chapter 6 Inorganic chemistry in the kiln system
109
6 Inorganic chemistry in the kiln system
As described in chapter 3, local reducing conditions caused by alternative fuel combustion in the
material inlet end of rotary kilns have the potential of increasing the internal circulation of sulfur,
chlorine and alkali metal species. In addition, alternative fuels may introduce increased amounts of
sulfur, alkali metals or chlorine to the system which will also affect the internal circulation
involving these elements and may subsequently affect process stability of the kiln system. This
chapter seeks to provide insight into the inorganic chemistry of the elements sulfur (S), chlorine
(Cl) and alkali metals (Na and K) in the kiln system, when alternative fuels are utilized. Main
emphasis is on sulfur, because this element is mostly affected by combustion conditions.
The chapter starts with a literature study about release and capture of inorganic volatile elements.
Then a section describes thermodynamic equilibrium calculations to theoretically investigate the
inorganic chemistry in the kiln system under different atmospheres.
The experimental part presents the results from experiments with sulfur release from calcined raw
materials in a laboratory scale experimental set-up under oxidizing and reducing atmospheres. And
finally, sulfur release from raw materials under the influence of different solid fuels is described.
The experiments with solid fuels are made in the specially designed high-temperature rotary drum
that was described in chapter 5.
6.1 Introduction
6.1.1 S, Cl, Na and K species in the rotary kiln, calciner and lower preheater cyclones
This section will briefly review the forms of the elements S, Cl, Na and K at different positions in
the kiln system.
Figure 6-1 shows the chemical bonding of main S-, Na-, K- and Cl-containing species at different
locations in the lower preheater cyclones, calciner, rotary kiln and clinker cooler. Multiple other
species are possible, but are omitted for clarity. The preheated raw material from the upper
preheater cyclones (not shown) enters the lower cyclone stages. Because the raw material have been
preheated to around 650C, only inorganic bound sulfur is present in the raw materials, mainly in
the form of alkali or calcium sulfates (Hansen, 2003). The alkali metals may be bound as sulfates,
chlorides or carbonates.
121
Chapter 6 Inorganic chemistry in the kiln system

110

Fuel and
primary air
(S, Na, K, Cl,
N2/O2)
Tertiary air
Calciner fuel
(S, Na, K, Cl)
Cooling air
(N2/O2)
Calciner
To preheater
Rotary kiln
~890
o
C
Coarse alternative
fuel particles
(S, Na, K, Cl)
Cement clinker
((Na, K)2SO4,
3K2SO4Na
2
SO
4
,
CaSO
4
, CaCl
2
,
K2SO42CaSO4)
~1450
o
C
~900-1000
o
C
~2000
o
C
Preheated raw
materials (~800
o
C)
((Na, K)2CO3, (Na, K)Cl, (Na, K)SO4,
CaCl2, CaSO4, CaCO3, CaO)
Partially calcined
raw materials
((Na, K)2CO3, (Na, K)Cl,
(Na, K)2SO4, CaCl2,
CaSO4, CaCO3, CaO)
~100
o
C
CaO + SO2 +
O2
CaSO4
Burning zone
Material inlet end
CaSO4 (Na,K)2SO4
(Na, K)Cl, SO2
SO2
SO2
From preheater (~650
o
C)
((Na, K)2CO3, (Na, K)Cl, (Na, K)SO4,
CaCl2, CaSO4, CaCO3)
(Na, K)Cl, CaCl2
(Na, K)Cl, Cl2
(Na, K)Cl, CaSO4

Figure 6-1: The identity of main S-, Na-, K- and Cl-containing species in the kiln system. Solid flows shown with solid
arrows and gas flows shown with dotted arrows.

S, Cl, Na and K may also be introduced to the kiln system with the fuels, depending on fuel type
(see also table 3-2 in chapter 3 with chemical compositions of relevant fuels):
TDF will break the C-S bonds in the original tire during pyrolysis at temperatures above 200C, and
the sulfur will be released to the gas phase as either H
2
S or SO
2
(Dez et al., 2004).
RDF consists of a variety of organic and inorganic compounds, and as a consequence, the inorganic
volatile input from RDF may fluctuate. S, Cl, Na and K in RDF tend to be released as HCl, alkali
chlorides, alkali sulfates, H
2
S and SO
2
(Kobyashi et al., 2005).
PVC decomposes in the temperature interval 200-400C to HCl and a coke-like residue
(Zevenhoven et al., 2002).
Meat and bone meal (MBM) devolatilizes at temperatures above 300C. Chlorine from MBM may
be partly devolatilized into the gas phase as HCl or alkali chlorides (Aho and Ferrer, 2005). The
sulfur from the MBM may be released to the gas phase as either SO
2
or H
2
S (Aylln et al., 2006). It
122
Chapter 6 Inorganic chemistry in the kiln system
111
is possible that the formed alkali chlorides and the gaseous SO
2
react to form alkali sulfates (Aho
and Ferrer, 2005).
The release of inorganic volatiles from straw, wood and annual biomass has been studied (Jensen et
al., 2000, Davidsson et al., 2002 and Knudsen et al., 2004). The general trend is that the release of
K, Cl and S is strongly dependent on both the temperature and the inorganic composition of the
fuel. In the temperature interval 200-400C the organic matrix is destroyed and Cl and K are
released from their original binding sites and transferred to a liquid tar phase. Chlorine may further
be released to the gas phase as HCl while potassium is bound as KCl, K
2
CO
3
or to char-matrix
functional groups such as carboxylics and phenoxides (Jensen et al., 2000). At temperatures above
500-700C, a large fraction of potassium in the form of KCl is released. The fuel sulfur is also
released from the organic S-compounds at these temperatures. At temperatures above 800C,
K
2
CO
3
decomposes whereas potassium is released as free K atoms or as KOH. At temperatures
above 1000C potassium may be released to the gas phase from the char matrix and from potassium
silicates (Jensen et al., 2000). Inorganic bound sulfur will also be released as SO
2
at these high
temperatures (Knudsen and Lin et al., 2004).
The calciner acts as a fast fluidized-bed reactor where good contact between hot gas from the rotary
kiln or tertiary air duct and raw materials from the preheater provide excellent conditions for gas-to-
solid heat transfer that drives the endothermic calcination of limestone. Besides calcination of
limestone, the conditions in the calciner with temperatures in the range of 800-900C, good gas-
solid mixing and excessive amounts of CaO also favor the SO
2
capture reaction (Lyngfelt and
Leckner, 1989):
CaO (s) + SO
2
(g) +O
2
(s) CaSO
4
(s) (6.1)
The relative amounts of alkali sulfates in the partially calcined raw material leaving the calciner are
small compared to the amount of CaSO
4
due to the excessive amounts of calcium (Jns, 2010). The
partially calcined raw material also contains a high concentration of chlorine in the form of either
alkali or calcium chlorides. This is due to the condensation of chlorine species from the kiln gases
when the gas is transported through the calciner and lower cyclone stages. Chlorides, which are
very sensitive to high temperatures, will to a high degree evaporate as alkali chlorides in the rotary
kiln, the evaporation factor typically being close to 1 (Locher et al., 2009).
Alkali metals may evaporate as chlorides, sulfates or elemental alkali metals. Complex vapour
equilibria exists in the rotary kiln, where alkali, sulfur and chlorine species may dissociate to
species that strongly interacts with combustion products such as CO
2
, CO, H
2
O, H
2
or O
2
(Bhatty et
al., 2004). As a result, the gas phase contains several other species than indicated in Figure 6-1, for
example SO
3
, HCl, NaOH and KOH.
During the gradual heating of the calcined raw material, a sulfate melt is formed which facilitates
formation of alkali sulfates, Na
2
SO
4
and K
2
SO
4
, which are more thermodynamically stable than
CaSO
4
(Choi et al., 1988). CaSO
4
will decompose at the high temperatures present in the burning
zone and predominantly form SO
2
:
CaSO
4
(s) SO
2
(g) + O
2
(g) + CaO (s) (6.2)
SO
2
formed in the rotary kiln is transported with the kiln gases to the calciner, where the previously
described sulfur capture on CaO takes place. Thus, an internal sulfur cycle in the rotary kiln and
123
Chapter 6 Inorganic chemistry in the kiln system
112
calciner is established. The fraction of sulfur released as SO
2
depends on the sulfur-to-alkali ratio in
the rotary kiln: High levels of alkali metals will favor formation of alkali sulfates rather than
formation of free SO
2
(Rosholm et al., 1998).
After the rotary kiln, the cement clinker is quench-cooled by ambient air as cooling air. The
relatively low chlorine content may be distributed in the major clinker phases or be present as
chlorides. Sulfur and alkali metals in the cement clinker are mainly bound as Na
2
SO
4
, K
2
SO
4
,
3K
2
SO
4
Na
2
SO
4
(aphthitalite), CaSO
4
, K
2
SO
4
2CaSO
4
(calcium langbeinite) or as substituents in
the major clinker phases, mainly alite and belite (Miller and Tang, 1996, Taylor, 1999, Twomey et
al., 2004).
The reason why CaSO
4
is sometimes found in clinker despite its thermodynamic instability at the
high temperatures that prevail in the burning zone, may be related to the nodulization environment:
The dense interior of large clinker nodules have been reported to contain elevated levels of both
alkali metals and sulfur, indicating that these inorganic volatiles may be trapped in the inner core of
clinker nodules (Masaki et al., 2002).
Experiments with raw materials mixed with gypsum have shown that an effective grinding to small
particle sizes will help to stabilize CaSO
4
(Tsoumeleas et al., 2000). This is achieved by the
formation of complex compounds such as 2CaSO
4
K
2
SO
4
and Na
2
Ca(SO
4
)
2
which can withstand
temperatures up to 1520C. This important stabilization may be explained by the fact that small
particles will be in better contact with each other in the molten phase of the clinker, and thereby
alkali and sulfur-containing species may easier form stable compounds.
6.2 Literature study on release and capture of inorganic volatile elements
6.2.1 High-temperature reactions between SO
2
and limestone
Absorption of SO
2
on limestone under fluidized bed and entrained flow conditions have been
studied extensively during the last decades (Lyngfelt and Leckner, 1989/1999, Dam-Johansen et al.,
1991 Hansen et al., 1991/1993, Hansen, 2003, Tarelho et al., 2005, Rasmussen, 2011). Even
though these investigations are not directly applicable to describe the conditions in the material inlet
end of cement rotary kilns, they do include relevant descriptions of mechanisms for sulfur release
and capture by limestone at temperatures up to 1200C and under oxidizing as well as reducing
atmospheres. An important point to note, however, is that these investigations only include
reactions between limestone and sulfur species, whereas a full investigation of the reactions taking
place in cement rotary kilns should also include other major components present in the cement raw
materials as they may affect the sulfur chemistry. In addition, the Ca/S molar ratio studied under
fluidized bed conditions were typically in the range of 1:1 to 4:1 which differs from the conditions
in cement rotary kilns where there is a massive excess of Ca relative to S: The Ca/S molar ratio in
the raw material that enters the rotary kiln is typically higher than 10:1 (Hewlett, 1998).
6.2.1.1 Mechanisms for sulfur capture and release from limestone
The following reaction mechanism for reductive decomposition of CaSO
4
in fluidized beds has
been proposed (Hansen et al., 1993):
CaSO
4
(s) + CO (g) CaSO
3
(s) + CO
2
(g) (6.3)
124
Chapter 6 Inorganic chemistry in the kiln system
113
CaSO
3
(s) CaO (s) SO
2
(g) (6.4)
CO is believed to react directly on CaSO
4
, giving CaSO
3
and CO
2
. CO
2
is rapidly desorbed, while
CaSO
3
disproportionates and forms CaO and SO
2
. CaSO
3
is believed to be an important reaction
intermediate. The formed CaO is subsequently sulphidized to CaS according to:
CaO (s) + SO
2
(g) + 3CO (g) CaS (s) 3CO
2
(g) (6.5)
Returning to oxidizing conditions, CaS is oxidized to CaO by the reaction:
CaS (s) + 1.5O
2
(g) CaO (s) SO
2
(g) (6.6)
The CaS oxidation is a rapid and very exothermal reaction which may lead to a temperature
increase (Tian and Guo, 2009). The final step is the formation of CaSO
4
by reaction of CaO, SO
2
and O
2
, which is also an exothermal reaction (Tian and Guo, 2009):
CaO (s) + SO
2
(g) + O
2
(g) CaSO
4
(s) (6.7)
Hansen et al. (1991 and 1993) studied phase equilibria for the SO
2
-CaO-CaSO
4
-CaS-CO-CO
2
system. They performed experiments in an electrically heated laboratory scale fixed bed reactor
developed to simulate the changing oxidizing and reducing conditions similar to the conditions
particles will experience in a fluidized bed reactor. They found that any transformation between
CaSO
4
and CaS takes place via CaO. This transformation cycle is shown in Figure 6-2, which also
illustrates the competition between sulfur capture and sulfur release. This competition depends on
parameters such as partial pressures of SO
2
, CO and CO
2
, see the phase diagram in Figure 6-3.
Temperature is a very important parameter: An increase in temperature in the interval 800-1200C
will shift the equilibrium towards CaO and release more SO
2
(Hansen et al., 1991).
CaO
CaO
Oxidizing conditions
CaSO
4
CaS
Reducing conditions
O
2
, SO
2
CO
SO
2
, CO
2
CO, SO
2
CO
2
O
2
SO
2
Figure 6-2: Sulfur transformation during periodically changing oxidizing and reducing conditions (Hansen et al., 1993).
125
Chapter 6 Inorganic chemistry in the kiln system

114


Figure 6-3: Phase diagram for the SO
2
-CaO-CaSO
4
-CaS-CO-CO
2
system (Based on thermochemical data from Barin,
1995).

Hansen et al. (1993) found that the competition between sulfur capture and sulfur release under
alternating oxidizing and reducing conditions were shifted towards more sulfur release when the
temperature increased.
Lyngfelt and Leckner (1989) reported a decrease in the sulfur capture ability of limestone at
temperatures above 880C and reducing conditions, due to reductive decomposition of CaSO
4
.
Tarelho et al. (2005) found a maximum sulfur capture efficiency of limestone at around 825C.
Higher temperatures reduced the sulfur capture efficiency, particularly under reducing conditions.

6.2.1.2 Effect of reducing agent on the sulfur release

During devolatilization and combustion of alternative fuels in an oxygen-lean environment several
reducing agents may be formed; CO, H
2
, CH
4
as well as other hydrocarbons. It is well documented
that reducing agents such as CO and H
2
can promote reductive decomposition of CaSO
4
(Hansen et
al., 1991 and 1993). The rate of reductive decomposition depends on the specific reducing agent.
Hansen et al. (1993) studied the rate of reductive decomposition with all three reducing agents CO,
H
2
and CH
4
. It was observed that the rate of decomposition was fastest with H
2
as reducing agent.
This is in agreement with earlier findings by Wheelock and Boylan (1960) who reported that the
rate was increased by a factor two or three when the reducing agent was H
2
instead of CO. Also
Diaz-Bossio et al. (1985) reported that the rate of reductive decomposition was fastest with H
2

relative to CO. Hansen et al. (1993) did not detect any reductive decomposition of CaSO
4
with CH
4

as a reducing agent even though the reaction is thermodynamically feasible. It was assumed to be
because reductive decomposition of CaSO
4
with CH
4
is a very slow process that did not have
sufficient time to proceed during the 30-second intervals used in the experiments.
Turkdogan and Vinters (1976) also showed that elemental carbon could be used as a reducing agent.
The conversion rate of CaSO
4
to SO
2
and/or CaS was reported to be determined by the rate of
oxidation of carbon to CO which then rapidly reduced CaSO
4
.

1,00E-05
1,00E-04
1,00E-03
1,00E-02
1,00E-01
1,00E+00
1,00E-04 1,00E-03 1,00E-02 1,00E-01 1,00E+00
P
a
r
t
i
a
l

p
r
e
s
s
u
r
e

o
f

S
O
2
[
a
t
m
]
Reducingpotential(p(CO)/p(CO
2
))
800C
900C
1000C
1100C
1200C
CaS
CaO
CaSO
4
126
Chapter 6 Inorganic chemistry in the kiln system
115
6.2.2 Removal of chlorine from the kiln system
Increased input from fuels or raw materials of chlorine in the kiln system has made it necessary for
many cement plants to install a so-called bypass system. The purpose of the bypass system is to
remove a fraction of the chlorine from the kiln gas stream, the primary aim being to avoid
blockages predominantly in the lower preheater cyclone stages, and to reduce kiln shell corrosion
and refractory degradation.
The bypass typically starts at a gas extraction point located immediately after the kiln gas outlet
from the rotary kiln. This position is usually chosen because the chlorine concentration in the flue
gas is high. A small bypass at this point is sufficient to remove a large amount of chlorine from the
kiln system (FLSmidth, 2008). The removal efficiency of chlorine depends on the chlorine
concentration in the kiln gas. It is usually sufficient to extract less than 10% of the kiln gas. Sutou et
al. (1999) reported that 90% of the chlorine can be removed by extracting 5% of the kiln exhaust
gas when the chlorine concentration in the calcined raw materials entering the rotary kiln is in the
range from 5000 ppm to 7500 ppm.
The extracted kiln gases containing chlorine and kiln dust are quench cooled by air to condense the
chlorides from the gas. The condensed chlorides are then separated from the gas in an electrostatic
precipitator (ESP) or a bag filter (BF). In some installations, a conditioning tower is positioned
between the quenching section and the ESP/BF in order to reduce the volume of gases and thereby
the size and cost of the ESP/BF. The gases from the ESP/BF are often drawn by their own fan to a
common stack. The separated dust has high chlorine content and may be reused in the cement plant,
used in other industries or disposed of (FLSmidth, 2008).
Besides achieving the above-mentioned advantages, the bypass reduces the flue gas flow through
the preheater and thereby reduces the load on the ID fan. The overall heat consumption generally
increases by 6-8 kJ/kg clinker per percentage point of bypass gas. Gaseous emissions of CO, HCl
and SO
2
may be emitted with the bypass gases unless the process is appropriately designed
(FLSmidth, 2008), see also section 3.4.2.
127
Chapter 6 Inorganic chemistry in the kiln system
116
6.3 Equilibrium calculations on influence of reducing conditions
To provide the basis for a better understanding of the inorganic chemistry in the rotary kiln material
inlet, thermodynamic equilibrium calculations were performed using the commercially available
software FactSage 6.0. This program uses the principle of minimization of the total Gibbs free
energy to calculate the equilibrium composition of a chemical system with known total elemental
composition, temperature and pressure.
6.3.1 Input data
The equilibrium calculations are based on the composition of calcined cement raw material and of
the flue gas leaving the rotary kiln. These compositions are shown in Table 6-1 and 6-2. It should be
noted that the calcined raw material used in the calculations contains a relatively high amount of
chlorine and sulfur, relative to the guidelines for a stable kiln operation.
Calcined raw material Weight-% Moles/kg clinker
SiO
2
18.00 3.000
Al
2
O
3
4.76 0.467
Fe
2
O
3
2.30 0.144
CaO 56.80 10.128
K
2
O 2.50 0.266
Na
2
O 0.20 0.032
S total 2.66 0.829
Cl 1.17 0.330
Table 6-1: Calcined raw material composition used in the equilibrium calculations.
Oxidizing
conditions,

Vol.-% Moles/kg
clinker
Reducing
conditions,

Vol.-% Moles/kg
clinker
CO
2
25.00 5.580 CO
2
25.00 5.580
H
2
O 5.00 1.116 H
2
O 5.00 1.116
O
2
5.00 1.116 O
2
0.00 0.000
CO 0.10 0.022 CO 2.00 0.446
SO
2
0.05 0.011 SO
2
0.05 0.011
N
2
64.85 14.475 N
2
67.95 15.167
Total 100.00 - Total 100.00 -
Table 6-2: Flue gas compositions used in the equilibrium calculations.
The following input data were used:
- Chemical compositions of calcined raw material and flue gas as specified in Table 6-1 and
6-2. All elements inserted as moles/kg clinker.
- It is a foregone conclusion that there is one kg calcined raw material and 0.5 Nm
3
flue gas
for each kg clinker produced.
- The following elements were included: C, H, N, O, S, Cl, K, Na, Ca, Si, Fe, Al.
- The Ilue gas is either oxidizing ( 1.3) or reducing ( 0.9).
- Temperature: 800-1500C, step size: 25C.
128
Chapter 6 Inorganic chemistry in the kiln system
117
- Pressure: 1 atm.
For the equilibrium calculations, ideal gas phase compounds, solid phase compounds and liquid
phase compounds were included from the FactSage 6.0 compound database. A list of all compounds
included in the equilibrium calculations are enclosed in Appendix D.
6.3.2 Limitations
The equilibrium model has the following limitations, which should be considered when comparing
the results to actual conditions:
- The model does not consider mass transfer and kinetic limitations. In an industrial rotary
kiln, the residence time may be too short for the species to come into contact with each other
and for reactions to take place. Furthermore, the residence time may be too short for the
transformation or decomposition of some compounds to take place. Therefore, the predicted
formation of compounds may exceed that occurring in reality.
- The model does not take into account in which form the elements are initially present in the
calcined raw material and flue gas. In reality, the elements may be bound in different
minerals and be released at different temperatures. This may affect the distribution of the
element between the solid, liquid and gas phase. It may also affect the compounds formed
and the phase distribution of other elements.
Because of these limitations, the results of the equilibrium calculations cannot be used for a
quantitative comparison to experimental data, but only to determine which species are
thermodynamically favored under different gas atmospheres.
6.3.3 Influence of reducing conditions
In order to investigate the effect of changing the atmosphere from oxidizing to reducing, the
calcined raw material is exposed to a) an atmosphere with 5 vol.-% oxygen and b) an atmosphere
with 0 vol.-% oxygen and 2 vol.-% CO. The exact composition of calcined raw material and flue
gas is given in Table 6-1 and Table 6-2.
The equilibria for oxidizing and reducing conditions are shown in Figure 6-4. Only the main sulfur-
containing compounds are included for clarity. The figure shows the sulfur distribution between
CaSO
4
, CaS and SO
2
at various temperatures and under either oxidizing or reducing atmosphere. It
is noted that no alkali sulfates are formed in the studied case, because formation of alkali chlorides
is thermodynamically favored.
The most important observation from Figure 6-4 is that the decomposition of CaSO
4
to CaO and
SO
2
takes place at lower temperatures when reducing conditions are prevailing: Under oxidizing
conditions, CaSO
4
is stable until approximately 1100C. At temperatures between 1100C and
1350C, CaSO
4
is gradually decomposed forming CaO and SO
2
. Above 1350C, CaSO
4
is not
thermodynamically stable and all sulfur is released as SO
2
. The situation is very different under
reducing conditions, where CaSO
4
is observed to be partly reductively decomposed already at
temperatures below 800C. The reductive decomposition continues until approximately 1300C. At
higher temperatures, all sulfur is present as SO
2
. The reductive decomposition of CaSO
4
takes place
129
Chapter 6 Inorganic chemistry in the kiln system
118
because sulfates can be used as an oxygen donor under reducing conditions. It should also be noted
that a fraction of the sulfur is present as CaS under reducing conditions. However, CaS is only
stable at temperatures below 850C, and the sulfur is released as SO
2
at higher temperatures. This is
in correspondence with the CaSO
4
-CaS-CaO-SO
2
-O
2
-CO phase diagram in Figure 6-3.
800 900 1000 1100 1200 1300 1400 1500
Temperature [C]
0
0.2
0.4
0.6
0.8
1
A
m
o
u
n
t
[
m
o
l
e
]
CaSO
4
(s) - = 1.3
SO
2
(g) - = 1.3
CaSO
4
(g) - = 0.9
SO
2
(g) - = 0.9
CaS (s) - = 0.9
Figure 6-4: S distribution in calcined raw material. Gas atmosphere is either oxidizing with 5 vol-% O
2
( 1.3) or
reducing with 2 vol- CO ( 0.9).
Through the equilibrium calculations it was also found that the phase distribution of Cl-containing
compounds such as HCl, NaCl and KCl is not affected by the change in gas atmosphere: It remains
the same whether the atmosphere is oxidizing or reducing. This result was also expected because Cl
is known to be much more sensitive to temperature than to gas atmosphere (Locher and Klein,
2009).
The sulfur retention in the calcined cement raw materials will depend on the specific degree of
reducing atmosphere. The equilibrium calculated sulfur retention under different atmospheres and
in the temperature interval 700-1100C is shown in Figure 6-5. This temperature interval is
representative for the conditions in the material inlet end of cement rotary kilns, where fully or
partly calcined raw materials enter the rotary kiln with a temperature of approximately 840-880C
and where the kiln flue gas leaves the rotary kiln with a temperature of approximately 1000-
1100C. It is observed that all sulfur is retained in the calcined raw material under oxidizing
conditions with 5-21% O
2
: These two curves are identical and at 100% sulfur retention in the
temperature interval. Also under inert conditions with 0% O
2
the sulfur retention is observed to be
100%, except above 1050C where a minor sulfur release is observed.
The sulfur release increases significantly when reducing conditions prevails. The sulfur release does
not increase with increasing concentration of the reducing agent CO. Instead, the situation can be
divided into mildly reducing conditions with 1-3% CO, and strongly reducing conditions with 4-
10% CO. The sulfur release is observed to be most significant under mildly reducing conditions,
where it accelerates at temperatures above 780C. At 1% CO, a fraction of the sulfur is released due
to reductive decomposition of CaSO
4
. But most of the sulfur is still retained in the calcined raw
130
Chapter 6 Inorganic chemistry in the kiln system
119
material. The thermodynamic calculations suggest that the sulfur is stabilized as K
2
SO
4
and
K
2
Ca
2
(SO
4
)
3
at 1% CO. 2% CO is observed to lead to a high sulfur release, reaching 50% at 950C.
This is because the compound K
2
Ca
2
(SO
4
)
3
is not stable under this reducing atmosphere, but at the
same time the reducing atmosphere is not strong enough to stabilize sulfur as CaS. Thus phase
equilibria are shifted towards free SO
2
. The sulfur release is lower at 3% CO because this stronger
reducing atmosphere favors formation of CaS. But the thermal stability of CaS decreases as the
temperature increases, so at 1050C the sulfur release reaches the same level as for 2% CO.
At strongly reducing conditions, 4-10% CO, the sulfur release is observed to be lower than under
mild reducing conditions. Again, this is because sulfur is retained in the calcined raw material as
CaS. But as the temperature increases and approaches 1100C the sulfur release increases rapidly
due to the thermal instability of CaS. These observations are in good correspondence with the phase
diagram shown in Figure 6-3, which indicates that sulfur may be stabilized as either CaSO
4
at a low
reducing potential or as CaS at a high reducing potential.
700 800 900 1000 1100
Temperature [C]
40
60
80
100
S
u
l
f
u
r
r
e
t
e
n
t
i
o
n
[
%
]
21% and 5% O
2
0% O
2
10% CO
5% CO
4% CO
3% CO
1% CO
2% CO
Figure 6-5: Sulfur retention in calcined raw material as a function of temperature and gas atmosphere. Note the cropped
Y-axis.
131
Chapter 6 Inorganic chemistry in the kiln system
120
6.3.4 Relative stability of sulfates towards reducing conditions
Sulfur introduced to the rotary kiln with the fully or partly calcined raw materials will mainly be in
the form of sulfates, CaSO
4
, Na
2
SO
4
and K
2
SO
4
. The distribution is dependent on parameters such
as concentrations of these individual elements, level of chlorine available to react with the alkali
metals, temperature, etc. Since it is clear that the stability of sulfates is affected by the gas
atmosphere, the relative stability of the sulfates CaSO
4
, Na
2
SO
4
and K
2
SO
4
has been studied
through equilibrium calculations.
Based on these equilibrium studies of the three dominating sulfates present in the kiln system, it is
clear that reducing conditions generally shifts the equilibrium towards more SO
2
in the gas phase.
The sulfates are used as an oxygen donor in the oxidation of CO to CO
2
.
An interesting observation was that the relative stability of the sulfates was Na
2
SO
4
> K
2
SO
4
>>
CaSO
4
, an observation that is confirmed experimentally. CaSO
4
is thus less stable and more critical
than the alkali sulfates. This observation may indicate that it is desirable to bind as much sulfur as
possible as alkali sulfates, in order to keep the sulfur evaporation in the kiln system as low as
possible. This has also previously been reported by e. g. Rosholm et al. (1998).
6.3.5 Concluding remarks about thermodynamic equilibrium calculations
Based on the thermodynamic equilibrium calculations it is clear that the stability of sulfates in the
inlet to the rotary kiln is strongly affected by the gas atmosphere. Especially CaSO
4
tend to
decompose at much lower temperatures under a reducing atmosphere. The alkali sulfates are also
affected by a reducing atmosphere, but not as significant as CaSO
4
.
The volatility of chlorine is not affected by the gas atmosphere, but is very sensitive to the
temperature.
The equilibrium calculations indicate that high levels of alkali will help to minimize SO
2
-release to
the gas phase, by forming stable alkali sulfates. However, this result is only valid under equilibrium
conditions, which are not necessarily present in the inlet to rotary kilns. It is possible that the alkali
content in the raw material will have no, or little, effect on the sulfur volatility in the inlet end of the
rotary kiln because the relative amount of calcium is much larger than the amount of alkali metals.
Thus, formation of CaSO
4
is much more likely than formation of the more stable alkali sulfates.
132
Chapter 6 Inorganic chemistry in the kiln system

121

6.4 Experimental
This section describes experiments made in order to improve the understanding of the inorganic
chemistry in the material inlet end of cement rotary kilns. The first part is about laboratory scale
experiments in a fixed bed tube furnace where SO
2
release from calcined raw material was studied
at different temperatures, atmospheres and time intervals. The second part describes SO
2
release
from raw materials under the influence of different solid fuels. These experiments with fuels are
made in the high-temperature rotary drum described in chapter 5, section 5.3.2.

6.4.1 Release of SO
2
from calcined raw material
The release of sulfur from cement raw materials as a function of gas atmosphere was quantified in
the temperature interval 900-1100C. Different gas atmospheres were tested but most of the
experiments were conducted with 5 vol.-% O
2
in N
2
to represent an oxidizing atmosphere and 2
vol.-% CO in N
2
to represent a reducing atmosphere. The temperature interval and gas atmospheres
resemble realistic temperatures and atmospheres in the material inlet end of cement rotary kilns.
The used cement raw materials were based on a mixture of pure chemicals of CaO, SiO
2
, Fe
2
O
3
,
Al
2
O
3
and a sulfate such as CaSO
4
, Na
2
SO
4
or K
2
SO
4
. These chemicals were mixed in the right
proportions to simulate a simplified calcined raw material, with a well-known content of sulfur. The
compounds are also representative for the form that the elements has on entry into the rotary kiln.

6.4.1.1 Laboratory-scale tube reactor set-up
A laboratory-scale tube reactor set-up was used for the experiments. The experimental set-up is
shown in Figure 6-6 and was found to be suitable for the experiments because it allowed a
continuous gas flow passing a raw material sample. The set-up consists of a gas mixing panel to
mix the gases in the desired concentration and flow, the electrically heated reactor in which the
sample is placed, filters to remove particles and moisture, a pump and the analyzers that measure
the gas concentration. Measured gas concentrations and temperatures are collected in a computer.


Figure 6-6: Tube reactor experimental set-up for investigation of SO
2
release from calcined cement raw materials.

The reactor depicted in Figure 6-7 is a two-zone electrically heated tube furnace in which a
horizontal alumina tube is mounted, having water-cooled flanges at both ends. The tube furnace has
133
Chapter 6 Inorganic chemistry in the kiln system
122
a maximum operating temperature of 1150C. The internal diameter of the inner tube is 47 mm and
it has a heated length of 900 mm. The sample is placed in a platinum/rhodium holder (95% Pt). The
dimensions of the sample holder is HxWxL = 2x3x10 cm, corresponding to a volume of
approximately 60 mL. During an experiment the sample holder is placed in the centre of the tube
furnace, and the tube furnace is sealed with a stainless steel plate which contains an opening for gas
inlet. The gas composition exiting the reactor is monitored on the computer. The sample will be
heated from room temperature and up to the desired temperature by conduction, convection and
radiation from the tube walls and gas.
Figure 6-7: Sketch of the tube furnace reactor with size specifications of the reactor tubes (D
O
= outer diameter, D
i
=
inner diameter, L = length).
In order to test whether the reactor set-up are producing stable and sufficient gas temperatures,
center-line temperature profiles in the tube reactor were established. It is concluded that, at the
position of the sample holder, the measured temperature corresponds fairly well with the reactor set
point temperature.
6.4.1.2 Method
The composition of used calcined cement raw material is shown in Table 6-3. Only the main
components, as well as a sulfur-containing component are included. The sulfur-containing
component is CaSO
4
, which is believed to be the dominant sulfur-containing component in calcined
raw material entering cement rotary kilns, due to the excessive amounts of calcium relative to alkali
metals. It should be noted that the sulfur content in the calcined raw material is relatively high (1.17
wt-%), but not unrealistic. Prior to the experiments, the synthetic calcined raw material was stored
under an inert atmosphere to prevent CaO from reacting with moisture or atmospheric CO
2
.
Component Amount (wt-%)
CaO 67
SiO
2
20
Al
2
O
3
5
Fe
2
O
3
3
CaSO
4
5
Table 6-3: Composition of calcined raw material used for the experiments.
The following procedure is followed during each experiment:
5 g of the calcined raw material sample is placed in a well distributed layer in the sample holder.
The sample amount is chosen to have a suitable amount of sulfur in the experiments in order to
134
Chapter 6 Inorganic chemistry in the kiln system
123
detect sulfur releases in the ppm range 0-1000 ppm. The sample container is then placed in the tube
furnace at the desired temperature. The tube furnace is sealed and the sample is heated for 10
minutes in an oxidizing atmosphere of 5% O
2
in N
2
. Practically no SO
2
release was observed under
the 10 minutes heating time indicating that CaSO
4
is stable under this oxidizing atmosphere. After
10 minutes, the gas atmosphere can be changed to the desired atmosphere whereby the experiment
begins. The gas flow is maintained at 10 NL/min at all times. The gas composition is logged for
subsequent evaluation of the sulfur release. In all experiments, the sulfur release was measured by
integrating the SO
2
-signal over time. The experiments always lasted 30 minutes, including 10
minutes heating time and 20 minutes experimental time.
In addition to the fixed gas atmosphere, experiments were also conducted with alternating gas
atmosphere between oxidizing and reducing. In these experiments, two-minute intervals were used:
2 minutes with reducing conditions (2 vol.-% of the reducing agent in nitrogen) followed by 2
minutes with oxidizing conditions (5 vol.-% O
2
in N
2
). This procedure was repeated over 20
minutes, which involved 5 reducing intervals and 5 oxidizing intervals.
6.4.1.3 Assumptions and uncertainties
This section discusses important assumptions and uncertainties that may influence the obtained
results.
1. It has been experimentally determined that the sample is heated for approximately 2-5
minutes in the tube furnace before the sample reaches the same temperature as the tube
furnace. The exact heating time depends on the sample size and temperature to be reached; a
5 g sample, which is the sample size used in the experiments, uses approximately 2 minutes
to reach 900C. In order to ensure sufficient time for sample heat up, the sample will stay 10
minutes in the furnace under oxidizing conditions before an experiment is started.
2. The gas passing through the reactor will be pure N
2
/O
2
/CO/H
2
/CH
4
in different
compositions. Under industrial scale conditions, the gas will also contain H
2
O, CO
2
, NO
x
,
SO
x
and various volatile species such as alkali metals or trace metals. It is a foregone
conclusion that the simplified gas composition used in the experiments contains the most
important components, and that the other components influence on the results can be
neglected.
3. The uncertainty on the O
2
, CO and CO
2
analyzers are +/- 2% of total span range of each
species while it is +/- 5% of total span range on the SO
2
measurement. These uncertainties
should be kept in mind when evaluating the results.
4. It is a foregone conclusion that all sulfur released from the sample will form SO
2
before
leaving the reactor. This assumption has been confirmed by two mass balances over two
different calcined raw material samples: After an experiment, the residual sulfur content of
the sample was determined by chemical analysis and compared to the sulfur amount that
was found in the flue gas as SO
2
. In the first case, the sulfur mass balance was found to be
93% and in the second case, 95% of the sulfur was found in either the sample or in the flue
gas as SO
2
. Considering the uncertainties with sample composition, precision of analyzers
and mass flow controllers, as well as the accuracy of the chemical analysis procedure, it is
135
Chapter 6 Inorganic chemistry in the kiln system
124
fairly safe to assume that nearly all released sulfur will be in the form of SO
2
. This
assumption is also confirmed by equilibrium calculations with the thermodynamics software
FactSage, which calculated that nearly all sulfur is present as SO
2
under the studied
conditions.
5. It is known from preliminary experiments that the physical shape of the sample holder and
sample mass will cause diffusion limitation of gas components such as CO to the sample.
However, since all experiments are made with similar sample masses and procedures, it is a
foregone conclusion that results will be comparable.
6. The flue gas composition will not change significantly in the distance between the reactor
and the analyzers. To minimize this uncertainty and prevent e.g. SO
2
absorption in filters
and condensers, these parts were regularly cleaned or replaced to minimize presence of
compounds able to absorb SO
2
. A gas with known content of SO
2
was also regularly sent
through the system to test if any SO
2
was absorbed before reaching the analyzers.
7. The calcined raw material to be used in the experiments is representative for calcined raw
material from an industrial plant. However, an industrial calcined raw material will contain
many minor components which could have an effect on results.
6.4.1.4 Results and Discussion
Figure 6-8 shows an example of the measured concentrations during an experiment with alternating
oxidizing and reducing conditions at 1100C. The first 10 minutes is the heating period under
oxidizing conditions, initially atmospheric air with 21% vol.-% O
2
followed by 4.75 vol.-% O
2
in
N
2
when the tube reactor is sealed. During the heating under oxidizing conditions, no SO
2
is
observed. After 10 minutes, the gas concentration is changed to 2 vol.-% CO in N
2
. Since the
change is rapid, some oxygen will be left in the reactor and will react with CO, forming CO
2
. Thus,
the CO
2
-concentration is observed to increase to approximately 1.5 vol.-%. After the last O
2
in the
reactor is consumed, the CO-concentration increases to approximately 1.8 vol.-%. At the same time,
a dramatic increase in the SO
2
-concentration is observed: From 0 ppm and up to 800 ppm. The
formed CO
2
is also observed at a level of around 0.2 vol.-%. The SO
2
-level drops relatively fast.
When the atmosphere is again changed to oxidizing (4.75 vol.-% O
2
), a new, smaller SO
2
-peak is
observed. The sulfur release cycle is observed to be repeated for each of the following
reducing/oxidizing intervals, with a slight decrease in the SO
2
each time, because the sulfur content
in the sample gradually decreases. The experiment is stopped after 30 minutes when the fifth
oxidizing interval is finished and the SO
2
-concentration has dropped to values close to 0 ppm.
However, it should be noted that more sulfur would be released from the sample if the cycles were
repeated: This was tried out and it can be concluded that the measured SO
2
-level gradually
decreases with time, but that it takes more than two hours before all sulfur has been released from
the sample.
136
Chapter 6 Inorganic chemistry in the kiln system
125
0 10 20 30
Time [min]
0
1
2
3
4
5
O
2
,
C
O
2
,
C
O
[
V
o
l
.
-
%
]
CO
CO2
O2
SO2
0
200
400
600
800
1000
S
O
2
[
p
p
m
V
]
Figure 6-8: Concentrations of SO
2
, O
2
, CO and CO
2
during an experiment with calcined raw material under alternating
oxidizing and reducing conditions. 1100C, 10 NL/min. 20 minutes of reaction time.
It is interesting to note from Figure 6-8 that the first SO
2
-peak during the reductive decomposition
of CaSO
4
only reaches 780 ppm while the next SO
2
-peak reaches 890 ppm. The first SO
2
-peak is
also wider than subsequent SO
2
-peaks. This tendency was observed in all experiments with
alternating reducing and oxidizing conditions. The same tendency has also been reported by other
researchers (Wheelock and Boylan, 1960, Ghardashkhani et al., 1989, and Hansen et al., 1991). The
reductive decomposition of CaSO
4
is initiated by an induction period which has been explained by
slow desorption of SO
2
and slow rate of nucleation of CaO. Ghardashkhani et al. (1989) reported
that shifting from reducing to oxidizing and back to reducing conditions during reductive
decomposition of CaSO
4
did not give rise to a new induction period. This is in agreement with the
experimental observations where the second SO
2
-peak reaches larger values than the first SO
2
-peak.
The characteristic SO
2
-peaks observed from Figure 6-8 during alternating reducing and oxidizing
conditions can be explained by the existence of phase equilibria between different sulfur species:
Under oxidizing conditions sulfur is stabilized in the form of CaSO
4
and under reducing conditions,
sulfur is partially stabilized as CaS. When the gas atmosphere changes, sulfur may be released to
the gas phase as SO
2
, as illustrated in Figure 6-2.
Release quantification
The release of sulfur during the experiments is calculated by integration of the gas concentration of
SO
2
over the relevant time interval:
2
6
Sulfur found
0
10
t
SO
PV
n y dt
RT

=
}
(6.8)
where y
SO2
is in ppmV and V is the gas flow in L/s. It should be noted that due to diffusion
limitation of gas components such as CO, only experiments made with similar sample masses can
be compared. The sum of sulfur found as SO
2
is divided by the total sulfur amount in the sample:
137
Chapter 6 Inorganic chemistry in the kiln system
126
sulfur found
sulfur in sample
S release 100%
n
n
= (6.9)
Effect of reducing atmosphere with CO
The sulfur release from calcined raw material has been studied at temperatures of 900C, 1000C
and 1100C. Two different atmospheres with 5% O
2
and 2% CO (pure nitrogen is the balance in
both cases) as well as alternating reducing and oxidizing conditions has been studied. The results
are shown in Figure 6-9. Under oxidizing conditions with 5% oxygen, the S release is
approximately 0% in the studied temperature interval. This result has been found to be valid at all
oxygen concentrations from 1% to 21%, indicating that CaSO
4
is generally stable under oxidizing
conditions and temperatures below 1100C. This finding is in agreement with the thermodynamic
equilibrium calculations which did not show any decomposition of CaSO
4
below 1150C.
Under reducing conditions with 2% CO, the sulfur release was still approximately 0% at 900C, but
increased to 17% at 1000C and 19% at 1100C. This significant increase in sulfur release under
reducing conditions may be explained by the formation of CaS from CaSO
4
in which CaO and free
SO
2
are reaction intermediates. Each experiment was repeated three times and the standard
deviation was found to be +/- 2%.
Under shifting oxidizing and reducing intervals with 2 minutes of 2%CO followed by 2 minutes of
5% O
2
continuously repeated 2 5 times, the sulfur release is observed to be significantly higher
than under constant reducing or oxidizing conditions: 14% at 900C, 40% at 1000C and 48% at
1100C. Each experiment has been repeated three times and the standard deviation were found to be
+/- 3%. This strong effect of alternating oxidizing and reducing conditions is in correspondence
with the sulfur transformation cycle shown in Figure 6-2: Under oxidizing conditions sulfur is
stabilized as CaSO
4
and under reducing conditions sulfur is stabilized as CaS. Every shift in
atmosphere leads to formation of CaO and free SO
2
. Formation of SO
2
is particularly large at higher
temperatures, whereas the lowest temperature of 900C can limit the sulfur release even under
alternating oxidizing and reducing conditions.
850 900 950 1000 1050 1100 1150
Temperature [C]
0
10
20
30
40
50
60
S
r
e
l
e
a
s
e
[
%
]
5 vol.-% O2
2 vol.% CO
Red./Ox. Intervals
Figure 6-9: Effect of gas atmosphere on the sulfur release from calcined raw material. Sulfur release evaluated after 20
minutes of reaction time.
138
Chapter 6 Inorganic chemistry in the kiln system
127
The results are in correspondence with previous work where sulfur release and capture have been
studied under alternating oxidizing and reducing conditions: Lyngfelt and Leckner (1989) reported
a decrease in the sulfur capture ability of limestone at temperatures above 880C and reducing
conditions, due to reductive decomposition of CaSO
4
. Hansen et al. (1993) found that the
competition between sulfur capture and sulfur release under alternating oxidizing and reducing
conditions were shifted towards more sulfur release when the temperature increased. Tarelho et al.
(2005) found a maximum sulfur capture efficiency of limestone at around 825C. Higher
temperatures reduced the sulfur capture efficiency, particularly under reducing conditions.
Effect of other reducing agents
Most investigations of reductive decomposition of CaSO
4
reported in literature use either CO or H
2
as reducing agent. Since combustion of alternative fuels in cement kilns may lead to formation of
significant amounts of CO, H
2
and hydrocarbons such as CH
4
under sub-stoichiometric conditions,
these three reducing agents have been investigated with respect to sulfur release from calcined raw
material in the temperature interval 900-1100C. The results are shown in Figure 6-10. The
experimental conditions were shifting oxidizing and reducing intervals with 2 minutes of the
reducing agent (2% of either CO, H
2
or CH
4
) followed by 2 minutes of 5% O
2
. The two-minute
intervals were continuously repeated 2 5 times, thus 20 minutes oI total reaction time. All
experiments were repeated three times with a standard deviation in the order of 3%. Under these
reaction conditions, the sulfur release when CO was used were 14%, 40% and 48% at 900C,
1000C and 1100C, respectively. When CH
4
was used, the sulfur release was 6%, 40% and 17% at
900C, 1000C and 1100C, respectively. And finally, when H
2
was used, the sulfur release was
8%, 27% and 20% at 900C, 1000C and 1100C, respectively.
850 900 950 1000 1050 1100 1150
Temperature [C]
0
10
20
30
40
50
60
S
r
e
l
e
a
s
e
[
%
]
5% O2 / 2% CO
5% O2 / 2% CH4
5% O2 / 2% H2
Figure 6-10: Effect of reducing agent on sulfur release from calcined raw material under alternating oxidizing and
reducing conditions. Sulfur release evaluated after 20 minutes of reaction time.
While the results with CO are observed to lead to increased sulfur release as a function of
temperature, the results with H
2
and CH
4
are more surprising: The sulfur release increases from
900C to 1000C but then decreases again at 1100C. This was unexpected since it was assumed
139
Chapter 6 Inorganic chemistry in the kiln system
128
that the sulfur release would increase with temperature regardless of the reducing agent. It was also
expected that H
2
would lead to a higher sulfur release than CO, and CH
4
would lead to a much
lower sulfur release than CO, according to Hansens findings: Hansen (1993) found that the rate of
reductive decomposition of CaSO
4
increased when H
2
was used instead of CO, and that CH
4
gave
no reductive decomposition at all. However, the reaction conditions used in Hansens experiments
were quite different from the ones used in this investigation, since he used relatively short time
intervals of 30 s / 30 s and only studied the effect of reducing agent at 850C. In this experiment,
longer time intervals of 2 min / 2 min were used and the temperature interval was 900-1100C. This
may explain why CH
4
is found to be of comparable efficiency as H
2
and CO in this investigation,
while Hansen reported it to be less efficient due to slower reaction kinetics.
Regarding the lower sulfur release at 1100C relative to 1000C when H
2
and CH
4
were used as
reducing agents, similar results have been reported by Kamphuis et al. (1993): At temperatures
above 950C, a fraction of the SO
2
was observed to be converted to H
2
S which was not detected by
the analysis system, when the flue gas contained H
2
. Based on a series of tests with known SO
2
concentrations in a H
2
/N
2
gas mixture passing through the reactor without a sample, Kamphuis et al.
established a correction factor to correlate for the formed H
2
S. It is likely that the same
phenomenon took place in the present experiments with H
2
and CH
4
since H
2
S can be formed from
both reducing agents. This will explain why the found sulfur release is lower at 1100C relative to
at 1000C. Prior to the experiments, thermodynamic equilibrium calculations had been made for the
calcined raw material exposed to the reducing agents H
2
and CH
4
. These thermodynamic
calculations indicated that practically all released sulfur would be present as SO
2
under the
experimental conditions. However, the thermodynamic calculations may be misleading since they
do not consider reaction kinetics: H
2
S may be formed as an intermediate species which leaves the
reactor before formation of SO
2
has occurred. A sulfur mass balance should be performed in order
to determine whether some sulfur has escaped from the system as H
2
S or not. Since a sulfur mass
balance has not been conducted for the experiments with H
2
and CH
4
it is not directly possible to
quantify the effect on sulfur release as a function of the reducing agent. However, the results do
indicate that all three reducing agents CO, CH
4
and H
2
have an effect on the sulfur release in the
studied temperature interval.
6.4.1.5 Conclusions for sulfur release from calcined raw materials
The release of sulfur from cement raw materials under both oxidizing and reducing conditions were
investigated. The investigated conditions resemble actual conditions in the material inlet end of
cement rotary kilns, where local reducing conditions may occur during combustion of solid,
alternative fuel particles and where solid/gas temperatures are usually between 900C and 1100C.
The following results were obtained:
- Experimentally, a clear tendency for increasing sulfur release was observed when the gas
atmosphere shifted from oxidizing to reducing: No sulfur release was observed under
oxidizing conditions (5 vol.-% O
2
), whereas the sulfur release increased to 19% under
reducing conditions (2 vol.-% CO) at 1100C. The sulfur release was particularly
significant if the calcined raw material was exposed to alternating oxidizing and reducing
conditions, which led to a sulfur release of 48% at 1100C.
- The sulfur release from calcined raw material was observed to be dependent on the
temperature: Under alternating oxidizing and reducing conditions, the sulfur release was
observed to be 14% at 900C and 48% at 1100C.
140
Chapter 6 Inorganic chemistry in the kiln system
129
- Sulfur release from calcined raw material was shown to take place in the presence of any of
the reducing agents CO, H
2
and CH
4
. However, it was not possible to directly quantify the
sulfur release as a function of the reducing agent, because part of the released sulfur from
the experiments with H
2
and CH
4
presumably formed H
2
S which could not be detected by
the analysis system.
6.4.2 Sulfur release from raw materials during combustion of solid fuels
The purpose of the experiments is to clarify how the sulfur release in the form of SO
2
is affected by
different solid fuels. The release of sulfur from cement raw materials as a function of fuel type and
fuel characteristics was quantified in the temperature interval from 700C to 1000C and at oxygen
concentrations from 5 vol.-% O
2
in N
2
to 21 vol.-% O
2
in N
2
. The temperature interval and oxygen
concentrations up to 10 vol.-% resemble realistic material temperatures and atmospheres in the
material inlet end of cement rotary kilns. The cement raw materials used were based on a synthetic
mixture of quartz sand and 5 wt.-% calcium sulfate, CaSO
4
. This represents a simplified batch of
cement raw material entering the rotary kiln, but with realistic sulfur content. Quartz sand was used
for practical reasons, as explained in section 5.4.1.
6.4.2.1 Experimental set-up and method
The experiments are made in the high-temperature rotary drum experimental set-up described in
chapter 5, section 5.3.2.
2.5 kg of raw material in the form of quartz sand with 5 wt.-% CaSO
4
is placed in the rotary drum.
This corresponds to a volumetric fill degree of 5%. The drum is rotated with 6 rotations per minute
in order to keep the raw material in a continuous rolling motion. The raw materials are then
preheated to the desired temperature, in the interval from 700C to 1000C. A constant flow of 100
NL/min of gas with the desired oxygen concentration is transported through the rotary drum and
will pass over the raw materials in the lower part of the rotary drum. The gas has been preheated to
the same temperature as the raw materials in order to obtain a uniform temperature at all positions.
The temperature and oxygen concentration are monitored continuously, and when these have
stabilized, a batch of solid fuel will rapidly be added into the raw materials in the rotary drum. SO
2
-,
O
2
-, CO- and CO
2
-concentrations are measured during the devolatilization and char combustion.
The gas composition is logged for subsequent evaluation of the sulfur release. In all experiments,
the sulfur release was measured by integrating the SO
2
-signal over time.
Since many of the fuels contain sulfur that will also be released during the experiments, baseline
experiments have been made, where the fuels are combusted in 2.5 kg of pure quartz sand, without
CaSO
4
. The measured SO
2
from the baseline experiments can only arise from the fuels and not from
the raw material. The amount of sulfur released during the baseline experiments is subtracted from
the sulfur released during the actual experiment in order to get the amount of sulfur that arises from
the raw material.
All experiments were repeated at least two times to ensure the repeatability. The sulfur release
values reported in the figures are the average values from the experiments. These values have been
corrected for the sulfur contribution from the fuels.
Five different solid fuels have been studied; tire rubber, pine wood, polypropylene, petcoke and
sewage sludge. In order to compare the sulfur release caused by each fuel with each other, the
amount of fuel is chosen to obtain the same energy input. Fuel proximate and ultimate analyses as
well as lower heating values are shown in Table 6-4. The lower heating values for the fuels are
141
Chapter 6 Inorganic chemistry in the kiln system
130
quite different. Thus quite different fuel sample amounts have been used in order to have the same
energy input in each experiment, regardless of the fuel. The energy input during the experiments
were chosen to be in the interval 15 to 50 kJ per batch, corresponding to fuel sample sizes from 0.3
g to 4.2 g. These energy inputs were chosen to obtain a suitable level of sulfur release, thereby
avoiding SO
2
-peaks exceeding the analyzer range.
Proximate analysis Ultimate analysis LHV
VM FC Ash C H N S MJ/kg
Tire rubber 64.6 32.6 2.8 87.4 7.1 0.3 1.2 36.9
Pine wood 75.3 24.5 0.2 38.9 5.2 0.1 - 16.2
Polypropylene 97.5 0.0 2.5 83.0 14.0 0.0 - 44.5
Petcoke 13.4 85.1 1.5 87.3 3.7 1.5 4.7 34.0
Sewage sludge 49.3 0.5 50.2 29.0 3.8 3.8 1.0 12.5
Table 6-4: Fuel analyses and lower heating values (LHV) for solid fuels used in the experiments. Units are in wt.-% as
received and MJ/kg.
Several different fuel particle sizes and shapes have been used in the experiments, and these are
presented in Table 6-5.
Fuel Shape Dimensions
Tire rubber granulate Irregular D
p
- 2 mm
Tire rubber cylinders Cylindrical L = 12 mm, D
p
= 9 mm
Pine wood cubes Rectangular Thickness - 10 mm
Pine wood saw dust Needles Thickness - 1 mm
Polypropylene flakes Rectangular Thickness - 1 mm
Petcoke Spherical D
p
- 1 mm
Sewage sludge Spherical D
p
- 0.5 mm
Table 6-5: Approximate shapes and dimensions of the fuels used in the experiments.
142
Chapter 6 Inorganic chemistry in the kiln system
131
6.4.2.2 Assumptions and uncertainties
This section discusses important assumptions and uncertainties that may influence the obtained
results.
1. The temperature at which the experiment is conducted is determined as the measured
temperature in the center of the rotary drum, above the raw material charge. It is a foregone
conclusion that the gas temperature and raw material temperature is the same at all positions
in the rotary drum. Temperature profiles have been measured at various positions in the
rotary drum, and the temperature was found to deviate no more than +/- 15C.
2. The experiments are conducted in the temperature interval from 700C to 1000C whereas
the raw material and gas temperatures in the material inlet end of modern industrial cement
rotary kilns are often in the interval 900C to 1200C. The temperature interval is thus only
partly representative for the industrial conditions. However, it is not possible to conduct
experiments above 1000C in the used experimental set-up.
3. The gas passing through the rotary drum will be pure N
2
and O
2
in different compositions.
During the devolatilization and char combustion, a fraction of the O
2
will be consumed, and
the ultimate combustion products H
2
O, NO, SO
2
, CO and CO
2
will be formed. In addition
some intermediate combustion products such as CH
4
or higher hydrocarbons, alkali species,
etc. may be formed. Under industrial scale conditions, the gas will also contain the
combustion products from the main kiln burner and may have an overall gas composition
different from the one used in these experiments. This difference in gas composition is
likely to affect the sulfur release, which is not only influenced by oxygen concentration but
also by the presence of reducing agents.
4. The uncertainty on the O
2
, CO and CO
2
analyzers are +/- 2% of span range for each species
while it is +/- 5%of span range on the SO
2
measurement. These uncertainties should be kept
in mind when the results are evaluated.
5. It is a foregone conclusion that all sulfur released from the sample will be oxidized to SO
2
before leaving the reactor. This assumption has been supported by equilibrium calculations
with the thermodynamics software FactSage, which calculated that nearly all sulfur is
present as SO
2
under the studied conditions.
6. The physical shape of the rotary drum reactor may cause diffusion limitation of oxygen to
the fuel particles. However, since all experiments are made under similar conditions, it is a
foregone conclousion that the results will be comparable.
7. The flue gas composition will not change significantly in the distance between the reactor
and the analyzers. To minimize this uncertainty and prevent e.g. SO
2
absorption in filters
and condensers, these parts were regularly cleaned or replaced to minimize presence of
compounds able to absorb SO
2
. A gas with known content of SO
2
was also regularly sent
through the system to test if any SO
2
was absorbed before reaching the analyzers.
143
Chapter 6 Inorganic chemistry in the kiln system
132
6.4.2.3 Results and discussion
General observations
Figure 6-11 shows an example of the flue gas composition during combustion of tire rubber
granulate at an initial oxygen concentration of 10 vol.-%. When the devolatilization starts, the O
2
-
concentration drops to approximately 7 vol.-%and the CO
2
-concentration increases to nearly 4 vol.-
%. The CO-concentration also increases shortly to around 600 ppm. The change in flue gas
composition is followed by a rapid increase in the SO
2
-concentration from 0 ppm to around 2,500
ppm. The SO
2
-concentration then gradually decreases towards 0 ppm again, as the CO/CO
2
-
concentrations gradually decrease.
0 20 40 60 80 100
Time [s]
0
2
4
6
8
10
O
2
a
n
d
C
O
2
c
o
n
c
e
n
t
r
a
t
i
o
n
s
[
V
o
l
.
-
%
]
0
500
1000
1500
2000
2500
3000
S
O
2
a
n
d
C
O
c
o
n
c
e
n
t
r
a
t
i
o
n
s
[
p
p
m
V
] O
2
SO
2
CO
2
CO
SO
2
from baseline experiment
Figure 6-11: Concentrations of O
2
, CO
2
, CO and SO
2
during an experiment with 0.81 g tire rubber granulate
corresponding to an energy input of 30 kJ. 900C, 5% fill, 6 rpm, 100 NL/min, 10% O
2
.
The sulfur release from the tire rubber and raw material is observed to begin almost immediately,
during the fuel heat-up and devolatilization. This is in correspondence with the work of Dez et al.
(2004) who reported that the C-S bonds in TDF will break at temperatures above 200C and form
either H
2
S or SO
2
. However, baseline experiments with fuel combustion in pure quartz sand without
CaSO
4
shows that only a minor fraction of the sulfur release arises from fuel-bound sulfur: Most of
the sulfur is released from the raw materials, which is also shown in Figure 6-11. The reason is
believed to be due to the formation of local reducing conditions near the raw material charge during
fuel devolatilization which leads to reductive decomposition of CaSO
4
. The tendency for a rapid
sulfur release during fuel devolatilization was observed for all experiments regardless of the fuel.
144
Chapter 6 Inorganic chemistry in the kiln system
133
Effect of energy input and fuel type
The sulfur release from the raw material as a function of the energy input, has been studied for five
different fuels at 900C and 10 vol.-% oxygen. The fuels were polypropylene flakes, petcoke, pine
wood saw dust, tire rubber granulate and sewage sludge, see Table 6-4 and Table 6-5 for details.
These fuels all have relatively similar average particle thicknesses in the interval 0.5-2.0 mm. The
results are shown in Figure 6-12. For all tested fuels it is observed that the sulfur release from the
raw materials increases as the energy input increases. It is also observed that the sulfur release is
quite different depending on the specific fuel: tire rubber granulate and sewage sludge leads to the
highest sulfur release while pine wood saw dust, petcoke and polypropylene all leads to a lower
sulfur release.
10 20 30 40 50
Energy input [kJ]
0
40
80
120
160
200
C
o
r
r
e
c
t
e
d
a
c
c
u
m
u
l
a
t
e
d
s
u
l
f
u
r
r
e
l
e
a
s
e
[
m
g
]
Poly propylene
Pet coke
Saw dust
Tire granulate
Sewage sludge
Figure 6-12: Corrected accumulated sulfur release as a function of energy input from different fuels. 900C, 10 vol.-%
O
2
, 5% fill, 100 NL/min, 6 rpm.
The reason why the sulfur release increase when the energy input increases becomes clear when the
flue gas compositions at each individual experiment are compared with each other: The larger the
energy input, the higher the rate of decrease of oxygen concentration will be during fuel
devolatilization, while both CO and CO
2
concentrations increase, thereby forming a higher degree
of reducing conditions near the raw material charge. In addition, local temperatures in the bed are
likely to increase when the energy input increases. Higher local temperatures may also cause an
increase in sulfur release, see section 6.4.2.3.3.
The large differences in sulfur release for identical energy inputs but by different fuels are
interesting. A detailed study of the flue gas concentration for each fuel type has not revealed any
clear trends about degree of reducing conditions and sulfur release. The reason is probably that the
exact particle sizes, shapes, densities, fuel combustion pathways and kinetics of the studied fuels are
quite different and that this also has an effect on sulfur release. The approximately spherical tire
rubber granulate and sewage sludge particles may for example lead to a high sulfur release because
of a relatively fast devolatilization combined with a high degree of mixing with the raw material.
The needle shaped saw dust particles gives a lower sulfur release than tire rubber granulate and
145
Chapter 6 Inorganic chemistry in the kiln system
134
sewage sludge despite the fact that the saw dust also has a fast devolatilization. An explanation
could be that the saw dust particles are less dense than the tire rubber and sewage sludge and that
the density difference leads to a low degree of mixing of saw dust into the raw material. The
polypropylene flakes have the lowest sulfur release which may be explained by a very fast
devolatilization; so fast that the polypropylene devolatilization ends before it has been well mixed
into the raw material. Finally, the small petcoke particles are a slow burning fuel with a volatile
content of only 13%. This leads to a relatively small decrease in oxygen concentration during the
fuel conversion compared to the other fuels and thereby a low degree of local reducing conditions
near the raw material bed.
Effect of temperature and fuel particle size
The sulfur release is studied as a function of the temperature and fuel particle size, which is
important for the devolatilization and char combustion times. Tire rubber and pine wood are used in
these investigations about particle size while the investigations about temperature also include
petcoke and polypropylene. The tire rubber is studied as one cylinder versus tire rubber granulate
and the pine wood is studied as one pine wood particle versus saw dust. The overall energy input is
the same in each experiment. Details about the fuels can be found in Table 6-4 and Table 6-5. The
results are shown in Figure 6-13 with sulfur release in mg versus temperature for the different fuel
particle sizes. It is observed that influence of the fuel particle size on the sulfur release at 700C and
800C is insignificant whereas the influence becomes more evident at 900C and, in particular, at
1000C: At 900C and 1000C the saw dust and tire granulate lead to a higher sulfur release than
for the wood particle and tire particle.
700 800 900 1000
Temperature [
o
C]
0
200
400
600
C
o
r
r
e
c
t
e
d
a
c
c
u
m
u
l
a
t
e
d
s
u
l
f
u
r
r
e
l
e
a
s
e
[
m
g
]
Tire granulate
Saw dust
Tire particle
Pet coke
Wood particle
Poly propylene
Figure 6-13: Corrected accumulated sulfur release from raw materials as a function of temperature and fuel particle
size. 5% fill degree, 10 vol.-% O
2
, 100 NL/min, 6 rpm. Energy input in each experiment: 30 kJ. The sulfur release from
the experiments with tire granulate and saw dust at 1000C has been estimated because the SO
2
-concentration exceeded
the analyzer range for these specific experiments.
The significant effect of temperature on the sulfur release was also to be expected based on the
thermodynamic equilibrium calculations (see section 6.3), due to the lower thermal stability of
146
Chapter 6 Inorganic chemistry in the kiln system
135
CaSO
4
at higher temperatures. In addition, a comparison of the flue gas compositions during the
combustion at different temperatures shows that the CO-concentration during fuel devolatilization
reaches higher values at higher temperatures than at lower temperatures. The reason is likely to be a
faster release of volatiles from the fuel particles at elevated temperatures which will increase the
tendency for local reducing conditions in the area around the fuel particles. This higher tendency for
local reducing conditions is likely to contribute to increase the sulfur release during the fuel
devolatilization.
The reason for the difference in sulfur release for different fuel particle sizes becomes clear when
the flue gas composition is studied for the individual experiments: The larger total surface areas of
the smaller fuel particles lead to a fast devolatilization and thus a fast release of volatiles. This leads
to significantly higher CO-concentrations and thus a higher reductive decomposition of CaSO
4
during devolatilization. An example is illustrated in Figure 6-14 which show flue gas compositions
during wood particle and saw dust combustion, respectively, at 1000C. At the wood particle
combustion, the CO-peak during devolatilization reaches 180 ppm and the SO
2
-peak reaches 1350
ppm. At the saw dust combustion, the CO-peak is significantly larger and reaches 27000 ppm (2.7
vol.-%) which leads to a SO
2
-peak well above the analyzer detection range of 3500 ppm. The
changes in O
2
-concentration are from the initial 10 vol.-% to 5 vol.-% for the saw dust and to 7 vol.-
% for the pine wood particle. Similar trends were observed for the experiments with one tire
particle versus tire granulate.
Since the experiments with saw dust and tire granulate at 1000C led to SO
2
-peaks well above the
SO
2
-analyzer range, it has been necessary to estimate the total sulfur release from these two
experiments. This induces some uncertainty on the exact amount of released sulfur in these two
cases at 1000C.
0 10 20 30 40 50 60 70
Time [s]
0
2
4
6
8
10
12
O
2
a
n
d
C
O
2
c
o
n
c
e
n
t
r
a
t
i
o
n
s
[
v
o
l
-
%
]
0
5000
10000
15000
20000
25000
30000
S
O
2
a
n
d
C
O
c
o
n
c
e
n
t
r
a
t
i
o
n
s
[
p
p
m
V
]
Saw dust
Wood particle
O
2
O
2
CO
CO
CO
2
CO
2
SO
2
SO
2
Figure 6-14: Flue gas composition during combustion of a pine wood particle with dimensions 30x15x10 mm and pine
wood saw dust with particle size < 1 mm. 10 vol.-% O
2
, 5% fill degree, 1000C, 100 NL/min, 6 rpm.
Figure 6-15 shows the sulfur release versus the total surface area of the fuel particles tire rubber and
pine wood at 900C. The sulfur release is observed to increase approximately linearly as a function
147
Chapter 6 Inorganic chemistry in the kiln system
136
of surface area for tire rubber. The sulfur release for pine wood is less significant than for tire
rubber but also may also be characterized as linear as a function of surface area. The large
difference in sulfur release between pine wood and tire rubber for the same surface area may be due
to factors such as devolatilization rate of the fuels and density. These factors are likely to affect to
what extent the raw materials will be exposed to local reducing conditions during fuel combustion.
But it is required to explore the reasons in greater details before any clear conclusions can be made.
0 50 100 150 200 250
Total fuel particle surface area [cm
2
]
0
40
80
120
160
200
C
o
r
r
e
c
t
e
d
a
c
c
u
m
u
l
a
t
e
d
s
u
l
f
u
r
r
e
l
e
a
s
e
[
m
g
]
Pine wood
Tire rubber
Figure 6-15: Corrected accumulated sulfur release versus the total surface area of pine wood and tire rubber fuel
particles. 900C, 10 vol.-% O
2
, 5% fill degree, 6 rpm, 100 NL/min.
148
Chapter 6 Inorganic chemistry in the kiln system
137
Effect of bulk oxygen concentration
The sulfur release has been studied as a function of the inlet oxygen concentration in the freeboard
gas above the raw material bed. The investigation has been conducted with polypropylene, petcoke,
pine wood cubes and tire rubber granulate, see Table 6-5 for details. The results are shown in Figure
6-16 with the sulfur release expressed in mg pure sulfur versus the oxygen concentration expressed
in vol.-%. It is observed that the sulfur release increases for all tested fuels when the oxygen
concentration decreases. Particularly the tire rubber granulate leads to an increase in sulfur release
of nearly a factor 8 when the oxygen concentration changes from 21 vol.-% O
2
to 5 vol.-% O
2
.
0 5 10 15 20
Inlet O
2
-concentration [Vol.-%]
0
50
100
150
200
250
C
o
r
r
e
c
t
e
d
a
c
c
u
m
u
l
a
t
e
d
s
u
l
f
u
r
r
e
l
e
a
s
e
[
m
g
]
Poly propylene
Pet coke
Tire granulate
Wood particle
Figure 6-16: Corrected accumulated sulfur release as a function of inlet oxygen concentration for four different fuels.
5% fill degree, 900C, 100 NL/min, 6 rpm. Energy input in each experiment: 30 kJ.
The reason for this high dependency of the oxygen concentration on the sulfur release becomes
clear when considering the changes in flue gas composition. Figure 6-17 shows the combustion of
tire rubber granulate at 21 vol.-% and 5 vol.-%, respectively. It is observed that the CO- and SO
2
-
concentrations reach approximately 500 ppm and 350 ppm, respectively, at 21 vol.-% O
2
, while the
CO- and SO
2
-concentrations reaches approximately 5700 ppm and 2900 ppm, respectively, at 5
vol.-% O
2
. Thus, the formation of the reducing agent CO is accompanied by a reductive
decomposition of CaSO
4
to SO
2
, and the degree of this reductive decomposition seems to be
directly related to the CO-concentration. The same trend was observed for the other tested fuels.
149
Chapter 6 Inorganic chemistry in the kiln system
138
0 25 50 75 100 125 150
Time [s]
0
5
10
15
20
O
2
a
n
d
C
O
2
c
o
n
c
e
n
t
r
a
t
i
o
n
s
[
v
o
l
-
%
]
0
1000
2000
3000
4000
5000
6000
S
O
2
a
n
d
C
O
c
o
n
c
e
n
t
r
a
t
i
o
n
s
[
p
p
m
V
]
21 vol.-% O2
5 vol.-% O2
O
2
O
2
CO
CO
CO
2
CO
2
SO
2
SO
2
Figure 6-17: Flue gas composition during combustion of 0.81 g (30 kJ) of tire rubber granulate at 5 and 21 vol.-% O
2
.
5% fill degree, 900C, 100 NL/min, 6 rpm.
6.4.2.4 Practical implications
The experiments with sulfur release during solid fuel combustion indicate that the sulfur release
from raw materials in the material inlet end of rotary kilns may be reduced by keeping the
temperature as low as possible and the oxygen concentration as high as possible, particularly during
fuel devolatilization. The results indicate that it may be advantageous to utilize relatively large fuel
particles because this will lead to longer fuel devolatilization times and thus reduce the tendency for
sudden formation of large CO-peaks followed by high sulfur release from the raw materials.
The experiments also gave an indication that the type of fuel was important for the extent of sulfur
release. The sequence was found to be tire rubber granulate > sewage sludge > pine wood saw dust
> petcoke > polypropylene flakes. However, it should be noted that the fuel particle size, and
possibly shape, is of high importance for the degree of sulfur release and other fuel particle sizes
may give quite different results.
6.4.2.5 Conclusions for experiments with sulfur release during fuel combustion
Sulfur release from cement raw materials during combustion of solid fuels has been studied
experimentally in a high-temperature rotary drum. The fuels were tire rubber, pine wood, petcoke,
sewage sludge and polypropylene. The sulfur release from the raw materials was observed to
increase when:
1. The inlet oxygen concentration decreased.
2. The temperature increased.
3. The total surface area of the fuel particles increased.
150
Chapter 6 Inorganic chemistry in the kiln system
139
These three parameters all had the potential of increasing the formation of local reducing conditions
in the raw material bed, as indicated by elevated levels of CO during fuel devolatilization. The
sulfur release is thus likely to increase as a consequence of the reducing conditions.
The type of fuel also had a significant effect on the sulfur release. The sequence was found to be tire
rubber granulate > sewage sludge > pine wood saw dust > petcoke > polypropylene flakes.
However, the sulfur release was significantly lower when the fuel was one tire rubber particle or
one wood particle, indicating the significant effect of fuel particle size and shape on the sulfur
release from the raw materials.
6.5 Conclusions for chapter 6
Volatile, inorganic elements such as sulfur, chlorine, potassium and sodium circulate to a high
degree in the rotary kiln, calciner and preheater due to evaporation-condensation reactions.
Chlorine, in particular, is a very volatile element with an evaporation factor close to 1. If utilization
of alternative fuels increases the input of particularly Cl it is often necessary to install a bypass to
remove a fraction of this volatile element, thereby preventing deposit build-ups.
The evaporation factor of sulfur depends on the amount of alkali metals available to form stable
alkali sulfates and/or alkali-calcium sulfates in the rotary kiln burning zone. Similarly, the
evaporation factor of alkali metals depends on the distribution between the stable sulfates and the
more volatile chlorides.
Local reducing conditions in the material inlet end of the rotary kiln have no direct effect on the
release of chlorine species to the gas phase: Chlorine release is to a very high degree influenced by
temperature and not by gas atmosphere.
By contrast, the release of sulfur is influenced by local reducing conditions in the material inlet end:
Sulfates present in the calcined raw material will be reductively decomposed by the presence of
reducing agents. Sulfur is mainly released as SO
2
. Thus, gas phase SO
2
-concentration increases
which may lead to increased deposit build-ups in rotary kiln, riser duct or lower cyclone stage.
The sulfur release was shown to be significantly higher under alternating reducing and oxidizing
conditions than under constant reducing conditions. It was also shown that sulfur release under
reducing conditions was particularly critical at high temperatures.
Experiments with sulfur release from raw materials during solid fuel combustion showed that the
fuel particle size was very important for the extent of sulfur release: Many small fuel particles led to
a higher sulfur release than few large fuel particles due to a faster devolatilization of the small fuel
particles.
The results indicate that problematic deposit build-ups caused by high SO
2
concentrations in the gas
phase may be reduced by minimizing the direct, physical contact between fuel particles and raw
material. Local reducing conditions in the raw material should be minimized by ensuring a high
excess air ratio in the rotary kiln and by optimizing the fuel-gas mixing.
Also, if temperatures in the material inlet end of rotary kilns can be kept relatively low, less sulfur is
expected to be released from the raw material during fuel devolatilization.
151
Chapter 6 Inorganic chemistry in the kiln system
140
6.6 References
Aho, M. and Ferrer, E.; Importance of coal ash composition in protecting the boiler against chlorine
deposition during combustion of chlorine-rich biomass, Fuel, 84, 201-212, 2005.
Aylln, M., Aznar, M., Snchez, J. L., Gea, G. and Arauzo, J.; Influence of temperature and heating
rate on the fixed bed pyrolysis of meat and bone material, Chem. Eng. Journal, 121 85-96, 2006.
Barin, I.; Thermochemical Data of Pure Substances. 3rd ed., VCH, Germany, 1995. ISBN: 3-527-
28745-0.
Bhatty, J. I., Miller, F. MacG. and Kosmatka, S. H.; Innovations in Portland Cement
Manufacturing, Portland Cement Association, Illinois, USA, Chapter 3.4, 2004. ISBN 0-89322-
235-1.
Choi, G.-S. and Glasser, F. P.; The sulphur cycle in cement kilns: Vapour pressures and solid-phase
stability of the sulphate phases. Cement and Concrete Research, Vol. 10, 367-374, 1988.
Dam-Johansen, K., Hansen, P. F. B. and stergaard, K.; High-temperature reaction between
sulphur dioxide and limestone III. A grain-micrograin model and its verification. Chem. Eng. Sci.,
Vol. 46, No. 3, 847-853, 1991.
Davidsson, K. O., Stojkova, B. J. and Pettersson, J. B. C.; Alkali emission from birchwood particles
during rapid pyrolysis, Energy & Fuels, 16, 1033-1039, 2002.
Diaz-Bossio, L. M., Squier, S. E. and Pulsifer, A. H.; Reductive decomposition of calcium sulfate
utilizing carbon monoxide and hydrogen, Chem. Eng. Sci., 40, 3, 319-324, 1985.
Dez, C., Martnez, O., Calvo, L. F., Cara, J. and Morn, A.; Pyrolysis of tyres. Influence of the
final temperature of the process on emissions and the calorific value of the products recovered,
Waste Management, 24, 463-469, 2004.
Farag, L. M. and Kamel, H. M.; Effect of high intakes of chlorine, sulfur and alkalies on cement
kiln operation. ZKG, 10, 586-590, 1994.
FLSmidth; Overcoming side effects of alternative fuels on pyro process. Highlights, 28, November
2008.
Ghardashkhani, S., Ljungstrm, E. and Lindquist, O.; Release of sulphur dioxide from calcium
sulphate under reducing conditions. 10
th
Int. Conf. of fluidized bed combustion, 611-615, 1989.
Hansen, P. F. B., Dam-Johansen, K., Bank, L. H. and stergaard, K; Sulphur retention on limestone
under fluidized bed combustion conditions An experimental study. Proceedings of the
international conference on fluidized bed combustion, 1 73-82, 1991.
152
Chapter 6 Inorganic chemistry in the kiln system
141
Hansen, P. F. B, Dam-Johansen, K. and stergaard, K.; High temperature reaction between sulphur
dioxide and limestone V. The effect of periodically changing oxidizing and reducing conditions.
Chem. Eng. Sci., Vol. 48, No. 7, 1325-1341 1993.
Hansen, J. P.; SO
2
Emissions from Cement Production. PhD Thesis, Technical University of
Denmark, Department of Chemical Engineering, 2003. ISBN 87-90142-96-9.
Hewlett, P. C. (Editor); Chemistry of Cement and Concrete; 4
th
edition, John Wiley & Sons Inc.,
New York, 1998. ISBN 0-340-56589-6.
FLSmidth; Overcoming side effects of alternative fuels on pyro process. Highlights, 28, FLSmidth,
Valby, Denmark, November 2008.
Jensen, P. A., Frandsen, F. J., Dam-Johansen, K. and Sander, B.; Experimental investigation of the
transformation and release to gas phase of potassium and chlorine during straw pyrolysis, Energy &
Fuels, 14, 1280-1285, 2000.
Jns, E.; Personal communication, FLSmidth, Valby, Denmark, January 2010.
Kamphuis, B., Potma, A. W., Prins, W. and Van Swaaij, W. P. M.; The reductive decomposition of
calcium sulphate I. Kinetics of the apparent solid-solid reaction. Chem. Eng. Sci., Vol. 48, 1, 105-
116, 1993.
Knudsen, J. N., Jensen, P. A. and Dam-Johansen, K.; Transformation and release to the gas phase of
Cl, K and S during combustion of annual biomass, Energy & Fuels, 18, 1385-1399, 2004.
Knudsen, J. N., Jensen, P. A., Lin, W., Frandsen, F. J. and Dam-Johansen, K.; Sulfur
transformations during thermal conversion of herbaceous biomass, Energy & Fuels, 18, 810-819,
2004.
Kobyashi, N., Itaya, Y., Piao, G., Mori, S., Kondo, M., Hamai, M. and Yamaguchi, M; The
behavior of flue gas from RDF combustion in a fluidized bed, Powder Technology, 151 87-95,
2005.
Locher, G. and Klein, H.; Modelling circulating sulfur, chlorine and alkali systems in the clinker
burning process; Part 2: Theory and discussion. Cement International, Vol. 7, 4, 64-75, 2009.
Lyngfelt, A. and Leckner, B.; Sulphur capture in fluidized bed boilers: The effect of reductive
decomposition of CaSO4. The Chem. Eng. Journal, 40, 59-69, 1989.
Lyngfelt, A. and Leckner, B.; SO
2
capture in fluidized-bed boilers: Re-emission of SO
2
due to
reduction of CaSO4. Chem. Eng. Sci., Vol. 44, No. 2, 207-213, 1989.
Lyngfelt, A. and Leckner, B.; Sulphur capture in circulating fluidized-bed boilers: Can the
efficiency be predicted? Chem. Eng. Sci., 54, 5573-5584, 1999.
153
Chapter 6 Inorganic chemistry in the kiln system
142
Masaki, K., Suzuki, M. and Maki, I.; Burning and nodulization process of clinker in the rotary kiln
as viewed from the fine textures of the constituent minerals. Cement and Concrete Research, 32,
1039-1044, 2002.
Miller, F. M. and Tang, F. J.; The distribution of sulphur in present-day clinkers of variable sulphur
content. Cement and Concrete Research, 26, No. 12, 1821-1829, 1996.
Rasmussen, M. H.; Low SO2 emission preheaters for cement production. PhD Thesis, Department
of Chemical and Biochemical Engineering, Technical University of Denmark, 2011.
Rosholm, P., Mortensen, A. H. and Hintsteiner, E. A.; Converting two kiln lines to 100% high
sulphur petroleum coke firing, ZKG International, February and April, 1998.
Sutou, K., Harada, H. and Ueno, N.; Concentration is the key, International Cement Review, 36-41,
June 1999.
Tarelho, L. A. C., Matos, M. A. A. and Pereira, F. J. M. A.; The influence of operational parameters
on SO
2
removal by limestone during fluidised bed coal combustion. Fuel Processing Technology,
86, 1385-1401 2005.
Taylor, H. F. W.; Distribution of sulfate between phases in Portland cement clinkers, Cement and
Concrete Research, 29, 1173-1179, 1999.
Tian, H. and Guo, Q.; Investigation into the behavior of reductive decomposition of calcium sulfate
by carbon monoxide in chemical-looping combustion. Ind. Eng. Chem. Res., 48, 5624-5632, 2009.
Tsoumeleas, C. I., Katsioti, M., Georgali, B. and Marinos, J.; Thermal stabilisation of CaSO
4
,
World Cement, 57-61 November, 2000.
Turkdogan, E. T. and Vinters, J. V.; Reduction of calcium sulphate by carbon, Trans. Inst. Min.
Metall. Sec. C, vol. 85, c113-123, 1976.
Twomey, C., Birkinshaw, C. and Breen, S.; The identity of the sulphur-containing phases present in
cement clinker manufactured using a high sulphur petroleum coke fuel. J. Chem. Technol.
Biotechnol., 79, 486-490, 2004.
Wheelock, T. D. and Boylan, D. R.; Reductive decomposition of gypsum by carbon monoxide. Ind.
Eng. Chem. 52, 3, 215-218, 1960.
Zevenhoven, R., Axelsen, E. P. and Hupa, M.; Pyrolysis of waste-derived fuel mixtures containing
PVC, Fuel, 81 507-510, 2002.
154
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
143
7 Analysis of industrial experience and combustion
model for industrial rotary kilns
The purpose of this chapter is to briefly describe and analyze industrial experience with
alternative fuel utilization in the material inlet end of rotary kilns. In addition, a full-scale model
for fuel combustion is suggested, based on the pilot-scale combustion experiments described in
chapter 5. Finally, the chapter gives a short evaluation of different alternative fuels regarding
their utilization in the material inlet end of cement rotary kilns.
The industrial experience, the full-scale model, and results from chapter 4 and 6 regarding
fuel/raw material mixing and sulfur release during fuel combustion may be used as a basis to
suggest how rotary kilns should be designed if it is desired to fire a fraction of the total energy
into the material inlet end while reducing negative impact on process stability or product quality.
7.1 Industrial experience with combustion of large fuel particles
This section summarizes observations and analyses that have been made for combustion of solid
alternative fuels in various FLSmidth rotary kiln systems. The aim is to provide insight into the
industrial scale experience with coarse alternative fuels fired into the rotary kiln, either directly
or indirectly through the calciner.
Initially, experience in wet kiln systems without preheater will briefly be described. Secondly,
Suspension Preheater (SP) kilns will be analyzed. It is relatively common to fire coarse
alternative fuels into the material inlet end of SP kilns. Finally, experience in modern calciner
kilns will be analyzed. Relatively little experience is available about alternative fuels firing into
the material inlet end of calciner kilns.
7.1.1 Wet process kilns
Wet process kilns are rarely constructed today due to a high energy consumption compared to
modern, dry process kiln systems. However, there are still many old wet kilns in operation, e.g.
at the Danish Aalborg Portland cement plant. The wet kilns are typically very long, with L/D
ratios around 40. The raw materials enter in one end of the rotary kiln as slurry with high water
content. All the processes of water evaporation, preheating, calcination and clinker reactions take
place within the long kiln.
Many wet process kilns (and long dry process kilns) uses mid-kiln firing of alternative fuels. The
fuel is typically tires. The fuel drops into the hot rotary kiln through a sluice which is installed in
the kiln shell. The technology has proven to work well on several long rotary kilns. Energy input
through mid-kiln firing may be in the order of 20% of the total energy fired (Hess, K. et al.,
2008). The technology has not been utilized in modern preheater or calciner kilns with shorter
rotary kilns, where firing in the calciner or in the material inlet end are preferred solutions.
155
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
144
7.1.2 SP kilns
The Suspension Preheater kiln (SP kiln) system consists of raw material preheating in 4-6
preheater cyclones prior to calcination and clinker reactions in the rotary kiln. SP rotary kilns are
relatively long because calcination takes place in the first section.
Many SP kilns fire 10-15% of their total energy input as coarse, solid fuels fed into the material
inlet end of the rotary kiln. This is a relatively high percentage, compared to calciner kilns, see
section 7.1.3. The reason why it works well is believed to be that the raw meal is only partly
calcined when it enters the rotary kiln typically 30-40% calcined. The heat developed under the
fuel combustion can therefore be directly used in the endothermic calcination process, thereby
efficiently controlling the local temperature and thereby the sulfur release.
The theoretically possible fuel quantity is normally limited to below 25% of total energy input
because of the limitation of excess air that can be passed through the burning zone. The typical
oxygen concentration in the material inlet end is quite low, about 2-3 vol.-%. Oxygen enrichment
in the material inlet end of the rotary kiln may be a feasible way to increase oxygen
concentration without having higher excess air ratios in the rotary kiln burning zone. This
approach has long been recognized, but has only become attractive in recent years due to the
availability of cheaper production methods of oxygen (Bhatty et al., 2004).
In practice, most SP kilns cannot fire more than 10-15% of their total energy input in the rotary
kiln material inlet end due to the risk of local reducing conditions and increasing temperatures
that may affect process and product stability. However, one European SP kiln produced by
FLSmidth is reported to be able to fire up to 35% of its energy input as whole tires in the
material inlet end of the rotary kiln. This extraordinary high percentage is due to low sulfur
content in the feed streams to the kiln system which minimizes problems in terms of deposit
build-ups. But it should be noted that a relatively large fraction of the production is reduced
clinker which have to be discarded. This specific cement factory accepts to discard a significant
part of the production because it have access to cheap alternative fuels in the form of whole tires
as well as low-sulfur waste fuels for firing through the kiln main burner (Rosholm, 2009).
Experiments made in a SP kiln with whole tires and tire chips indicate that whole tires are more
attractive than tire chips: The tendency to deposit build-ups and reduced clinker was less severe
with whole tires relative to the situation with tire chips (Jannerup, 2009). These results indicate
that the smaller tire chips can more easily be buried in the material bed and create local reducing
conditions. The larger whole tires will to a larger degree stay above the bed and be exposed to
higher temperatures as well as to higher oxygen concentrations from the freeboard gas. These
results have to the authors knowledge never been published but they are important since they
prove that there is no reason to shred the tires before firing them into the rotary kiln.
Furthermore, the results are in fine correspondence with the results from the mixing experiments
presented in chapter 4, where it was also observed that large fuel particles maintained a better
contact with the gas above the raw material bed than smaller fuel particles.
156
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
145
7.1.3 Calciner kilns
In modern cement plants, raw meal is preheated to calcination temperature in a multi-stage
cyclone preheater and most of the calcination process takes place in a separately fired calciner.
The In-Line Calciner (ILC) is an example of such a modern kiln system. Alternative fuels are
today widely used in calciner kilns. But it is most common to fire the fuels in the traditional
firing points in the calciner and main burner. Firing with coarse alternative fuels directly into the
rotary kiln is generally restricted to small percentages of the total energy input, due to low
oxygen concentrations about 2-7 vol.-% and high calcination degrees above 90% of raw meal
entering into the kiln. These conditions create a high risk for local reducing conditions in the
material charge and/or increasing temperatures in the rotary kiln material inlet or riser pipe.
In the following, full-scale experience with alternative fuel firing in the material inlet end of
calciner kilns at eight different cement plants will be described.
Plant 1
This cement factory is of the ILC type pyrolysis tires in a separate gasifier, whereafter the tire
char is introduced into the kiln riser duct and, subsequently, into the rotary kiln material inlet.
Besides tire char coal and petcoke are fired in the calciner and main burner. Plastic waste is also
fired in the main burner. Despite a relatively high O
2
-concentration in the kiln material inlet of 4-
6 vol.-%, it is clear from flue gas measurements that the SO
2
-concentration increases
significantly when tire char is fired into the kiln material inlet. This is illustrated in Figure 7-1.
When neither tire char nor plastic is fired in the kiln system, the SO
2
-concentration in the flue
gas is around 1800 ppm. When plastic is co-fired through the burner along with coal and
petcoke, the SO
2
-concentration increases to 3000-5000 ppm. Also, small CO peaks are seen. And
when both plastic and tire char is fired simultaneously, the SO
2
-concentration increases above
10000 ppm, which is the detection limit for the analyzer. The oxygen concentration fluctuates
around 4-6 vol.-% most of the time. However, a sudden drop in the oxygen-concentration is
observed with intervals, simultaneously with increasing CO-concentrations. This indicates that
there are some irregularities in the alternative fuel combustion, and that local reducing conditions
have been present in the material charge. This type of situation with CO peaks and increasing
SO
2
-concentrations is often reported from cement plants that fire solid fuels into the material
inlet end of the rotary kiln.
157
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
146
Figure 7-1: Flue gas analysis in kiln material inlet (Jns, 2007).
Plant 2
This cement plant of the ILC-type fires approximately 10% of the total energy consumption as
whole tires fired into the material inlet end of the rotary kiln. This is a relatively high energy
percentage for an ILC kiln, but despite this, none of the typical problems associated with local
reducing conditions have been observed. The remaining 90% of the energy is supplied from
heavy fuel oil fired in the main burner and calciner. The ILC kiln is similar to other ILC kilns in
most ways, so it is not directly possible to point out a particular reason why this plant is able to
fire a large percentage of fuel in the kiln material inlet end. But several possibilities exist: The
rotary kiln has a small length-to-diameter ratio of 11. The kiln inner diameter is 5 m which is
relatively large for a typical clinker production around 3200 tpd. This means that the material
filling degree is only around 6%. The combination of a low material filling degree and a large
fuel particle in the form of a whole tire is likely to provide good fuel-air contact in the rotary
kiln. Thus, the tendency for local reducing conditions in the material charge may be minimized.
Another possibility may be that the whole tires are delivered continuously, on a kg-per-hour
basis. Thus, fluctuations in energy input to the rotary kiln can be minimized. A third possibility
may be the rather unusual main fuel, heavy fuel oil. Oil consists of hydrocarbons and during the
combustion process it will produce more water vapour than coal. Thus the flue gas in the rotary
kiln will contain high water content. It is well known that OH radicals from water accelerate the
oxidation of CO (Turns, 2006). This phenomenon could consequently decrease the degree of
local reducing conditions in the material charge.
158
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
147
Plant 3
Another ILC rotary kiln fires 15-20% of the total energy as tire chips into the middle of the
calciner, see Figure 7-2. The tire chips are in sizes of up to 10 cm in length and width. It is
estimated that 10-20% of the tire chips falls through the kiln riser duct and into the kiln material
inlet end where the combustion is completed (Green-Andersen, 2008). The kiln operation is not
reportedly affected by these tire chips in the form of local reducing problems in the kiln inlet.
But another challenge is formation of so-called bird nests in the kiln riser duct: Steel wires
from the tire chips get stuck in the riser duct and dust transferred with the kiln gas accumulates in
the network of steel wires. This creates deposits in the kiln riser duct that have to be manually
removed. At this particular plant, the main challenge associated with firing tires in the calciner,
riser duct and kiln inlet is thus the steel wires from the tires and not local reducing conditions.
Figure 7-2: ILC kiln with TDF chip firing in the middle of the calciner. The TDF chips drop into the kiln inlet
before reaching full conversion.
Plant 4
This cement plant of the SLC-D (Separate Line Calciner Downdraft) type has experience with
firing of many different alternative fuels: The plant has during the last 10 years fired different
hazardous solid and liquid waste fuels such as pesticides, organic solvents, cyanide and during
the last three years, whole tires (Paone, 2009). The fuels are fired into the lower part of the
preheater tower and are fully or partly combusted before reaching the rotary kiln material inlet.
The plant has due to the various alternative fuels experienced substantial fluctuations in the input
levels of sulfur and chlorine, as well as other elements. This has been compensated for by mixing
different fuel types in order to keep a fairly constant input level of different elements. Experience
from the plant is that the energy percentage in this position should not exceed 10% of total
energy fired in the plant. The remaining 90% energy is fired in either the kiln main burner or in
the calciner.
159
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
148
Plant 5
This cement plant of the ILC type has experience with firing of 1-2 energy-% of the alternative
fuel criolita into the kiln riser pipe: Criolita is a waste product from the aluminium producer
Alcoa where it is used as an electrode in aluminium production. It is in addition to the high
carbon content, a potential source of fluor which is an important mineralizer in cement
production. However, during trial tests it was observed that firing of criolita in the kiln riser pipe
caused increased formation of gypsum-rings in the kiln inlet. The same tendency had been
observed with previous tests with firing of petcoke in the kiln riser pipe. The tests were cancelled
and it was decided not to fire criolita in the kiln riser pipe due to the risk of kiln rings. It should
be noted that the kiln operation was generally unstable with low oxygen and NO concentrations
in the kiln inlet, as well as frequent CO peaks (Nielsen and Jns, 2000). It is possible that a
higher excess air ratio in the kiln inlet would be sufficient to avoid increased formation of kiln
rings when fuels are combusted in the kiln riser duct.
Plant 6
This ILC kiln has a production capacity of 3000 tpd and uses several different fuels: Petcoke,
Liquid Waste Derived Fuel (LWDF), whole tires and shredded solid industrial waste. The plant
was able to fire 1 tph of whole tires into the lower calciner part. Here the whole tires were held
by a fork in the hot gas stream from the tertiary air duct and rotary kiln before being pushed into
the kiln riser duct and kiln inlet. However, after installation of a chloride by-pass, the process
conditions in the kiln riser duct changed with less dust entrained in the flue gas and gas
temperatures increasing from approximately 1000C to 1050-1100C (Lewis and Shenk, 2009).
This led to build up problems in the kiln riser duct unless the tire feeding was reduced to 0.5 tph.
Since whole tires were the cheapest fuel in the cement plant it was desirable to maintain the feed
rate at 1 tph. Attempts were made to increase the oxygen concentration in the kiln material inlet
from 3-3.5% to 6-7% and install additional air blasters in the kiln riser duct. These two
modifications were proven to be successful in removing build ups in the kiln riser duct. The
overall pressure drop over the preheater tower was noted to decrease due to less build up in the
riser duct. Consequently, the plant is able to run at higher production capacity as a result of
increased draft capacity.
Plant 7
Another cement plant of the ILC-type utilizes both whole tires and 300x300 mm tire chips in the
material inlet end of the rotary kiln (Lund, 2010). The experience from this cement plant is that
whole tires has less negative impact on process stability, presumably because the tire chips
devolatilize too rapidly due to their higher surface area. This leads to high CO concentrations in
the material inlet end. The cement plant still accepts to receive tire chips because transportation
costs is cheaper: Tire chips can be packed better than whole tires, making it possible to transport
more on each truck. When whole tires are used, the capacity is around 3.6 t/h which corresponds
to 12-14% of the total energy requirement. This amount of tires can replace 3-5 t/h of coal.
However, it is only possible to utilize whole tires some of the time due to deposit build-up
formation in the riser duct, which must be manually removed. The daily average energy supplied
by tires are around 7% of the total energy requirement (Lund, 2010).
160
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
149
Plant 8
This plant of the ILC type has a production capacity of 3300 tpd in a rotary kiln with a length of
69 m and a diameter of 4.6 m, corresponding to an L/D ratio of 15 (Valchev, 2010). The cement
plant utilizes a variety of alternative fuels, mainly solid refuse derived fuels, whole tires, meat
and bone meal and flower shells. The average alternative fuel share is 51-55% of the total energy
requirement. Approximately 3 t/h of whole tires and 3 t/h of meat and bone meal is fed to the
material inlet end of the rotary kiln, and occasionally automobile interior components and waste
wood are also fed to the material inlet (Valchev, 2010). The cement plant has experienced
several problems with deposit build-ups, but have optimized the process stability by installation
of a chlorine bypass. However, it is still necessary to remove deposits in bottom stage cyclones,
riser duct and material inlet of the rotary kiln. Cleaning is done with cardox and air blasters as
well as by manual cleaning 6 to 10 times a day by one or two workers. Manual cleaning is done
with long metal pipes and pressurized air (Valchev, 2010).
7.1.4 Conclusions for industrial experience
This section summarized experience with combustion of large solid fuel particles in the material
inlet end of rotary kilns. Firing of fuels into the kiln material inlet end is most common in long
wet or dry rotary kilns, or in SP kilns where it is common to feed 10-15% of the total energy in
this position.
In modern calciner kilns (ILC or SLC kilns) it is generally regarded as problematic to feed solid
fuels into the kiln material inlet end. However, some calciner kilns do feed up to 10% of the total
energy in the kiln material inlet end. Experience has shown that higher feed ratios may result in
local reducing conditions, and consequently CO peaks, increased SO
2
-concentration in the rotary
kiln exit gas and heavier deposit formations in the kiln and riser duct.
The reason why long wet and dry rotary kilns and SP kilns are believed to be better suited for
firing in the kiln material inlet end is because the raw material in the kiln material inlet end is
only partly calcined. Thus, the raw material may be able to take up heat from the fuel
combustion so that high temperatures are minimized, thereby minimizing sulfur release. Another
theory is that during the calcination, CO
2
is released from the raw material and this creates a
bubbling, dynamic environment similar to that in the calciner. This environment promotes heat
and mass transfer between the material charge and the freeboard gas, thereby minimizing risk of
local reducing conditions in the material charge.
Most experience with alternative fuel utilization in the material inlet end is based on combustion
of tires which is a very common alternative fuel in the cement industry. It is possible that solid
fuels with different volatile/char contents and fuel conversion pathways can be fed into the kiln
material inlet end in higher ratios due to less impact on the inorganic chemistry.
It is interesting to note that experience with higher oxygen concentrations in the kiln inlet and
lower material filling degrees in the rotary kiln seems to have a positive effect on the process
stability during combustion of fuels in the kiln material inlet end. Experience with tires in two
different rotary kilns also indicates that the particle size is of importance: Whole tires seem to
have less negative impact on process stability than tire chips which become buried in the bed.
161
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns

150

7.2 Modeling combustion of solid fuels in industrial rotary kilns

The drying, devolatilization and char oxidation models developed in chapter 5 is now used to
estimate conversion times of solid tire particles in industrial rotary kilns. Tires are used in the
model because there is industrial experience with tire combustion, as described in section 7.1.
Thus it is possible to compare model results with industrial experience. But the same model
approach may be used for other fuels as well. The aims of these simulations are to:

a) Predict drying, devolatilization and char oxidation times for the fuels during their
transport into the rotary kiln.
b) Predict oxygen consumption in the rotary kiln during fuel conversion in the material inlet
end.
c) Predict approximate gas and bed temperatures during fuel conversion in the material inlet
end.

These model predictions may be useful in determining how large quantities of fuels that can be
accepted in the material inlet of the rotary kiln based on fuel characteristics and operational
parameters.

Figure 7-3 illustrates qualitatively how deep into the rotary kiln solid fuels will reach during
drying, devolatilization and char oxidation. In addition, the figure illustrates how temperature
and oxygen concentrations with and without fuel combustion in the material inlet end may be. As
shown on the figure, the temperature of both gas and bed will increase with solid fuels entering
the material inlet. And the oxygen concentration will decrease to a lower level in the material
inlet end of the rotary kiln.


Figure 7-3: Temperature and oxygen concentration profiles in a rotary kiln with and without alternative fuel
utilization in the material inlet end.






Gas flow
Solid flow
Rotary kiln
Kiln main burner
O2, out 21 vol.-%
O
2,in
2-7 vol.-%
(without alternative fuels)
T
g,in
1100C
(without alternative fuels)
T
g,in
with alternative fuels
T
g, max
2000C
Drying and
devolatilization zone
Char oxidation zone
O2,in with alternative fuels
Tg, out Tg, max
T
b, in
900C
Tb, out 1450C
162
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
151
Before the temperature, oxygen and fuel conversion models are developed, the following main
assumptions are made:
1) The oxygen profile in the gas phase is assumed to be evenly distributed over the radial
cross section.
2) Heat effects of calcinations reactions in the rotary kiln inlet are neglected.
3) Heat transfer through the kiln wall is neglected.
4) The gas and bed temperatures are modelled by a simple linear correlation. Conductive,
convective and radiative heat exchange is assumed contained in this correlation.
At conditions pertinent to the material inlet of rotary kilns, the oxygen concentration is expected
to be more important for fuel conversion than temperature. Consequently, the relatively simple
assumptions regarding the energy balance are considered to provide satisfactory model results in
terms of point a) and b) above.
The models for gas and bed temperatures along the kiln length and during fuel combustion are
included in Appendix E. Stoichiometric calculations (see Appendix F) coupled with the fuel
conversion profiles are used to determine the axial oxygen profile.
The used dimensions and values are summarized in Appendix B and G. The calculations are
based on a modern FLSmidth ROTAX 2 kiln with a clinker production of 3,500 tpd.
163
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns

152

7.2.1 Selected full scale model results
Figure 7-4 shows the estimated degree of conversion of tire-derived fuel (TDF) fed to the
material inlet end of a rotary kiln at a rate of 0.076 kg/s corresponding to around 2% of the total
thermal energy input necessary to produce cement clinker. The time for devolatilization is fast,
around 30 s, compared to the time for char oxidation which is around 1600 s. In the studied case,
this corresponds to 2 m of kiln length for the devolatilization and 56 m for the char conversion.
Thus 58 m of kiln length is required for the total conversion of the TDF. The long conversion
time for the char is due to mass transfer limitations of oxygen to the fuel, both from the bulk gas
to the bed surface and from the bed surface into the active layer of the bed. It may be shown that
the conversion time is roughly halved if the oxygen concentration is doubled, or if the fuel
amount is halved.

Figure 7-4: Degree of conversion for TDF as a function of kiln length in the material inlet end of a 60 m long rotary
kiln. 5 vol.% O
2
in kiln burner flue gas. Tire rubber average particle diameter = 11 mm. Fuel amount: 0.076 kg/s
corresponding to 2% of the total energy input to the kiln system.

Figure 7-5 shows the gas and bed temperature profiles through the rotary kiln during the tire char
combustion. The gas temperature is observed to decrease linearly from an average gas
temperature of 1600C in the burning zone to approximately 2 m from the material inlet end.
Here, the fuel devolatilization begins, thereby changing the slope of the temperature profile due
to the heat release from the volatile combustion. Similarly, the bed temperature is observed to
increase fastest from 900C in the first two meters of the material inlet and then increase at a
slower rate during the rest of the rotary kiln.
164
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
153
Figure 7-5: Temperature profiles through rotary kiln during combustion of TDF in a 60 m long rotary kiln. 5 vol.-%
O
2
in kiln burner flue gas. Tire rubber average particle diameter = 11 mm. Fuel amount: 0.076 kg/s corresponding to
2% of the total energy input to the kiln system.
Figure 7-6 shows the changes in bulk gas oxygen concentration through the rotary kiln when
TDF is combusted in the material inlet end of the rotary kiln. The oxygen concentration in the
flue gas from the kiln burner is 5 vol.-% corresponding to an excess air ratio of 1.33. This
oxygen concentration decreases slightly to 4.8 vol.-% during the char oxidation. And then it
decreases fast to 4.3 vol.-% during fuel devolatilization. Overall the change in oxygen
concentration is around 0.7 vol.-% through the rotary kiln. This indicates that the slow char
oxidation is not only due to lack of oxygen but also mass transfer of oxygen to the char: Most of
the oxygen does not reach the char before leaving the rotary kiln. If the rotary kiln is longer, the
fuel amount can be increased because the longer retention time will ensure more time for char
oxidation.
Figure 7-6: Oxygen fraction in bulk gas through rotary kiln during combustion of TDF in a 60 m long rotary kiln. 5
vol.-% O
2
in kiln burner flue gas. Tire rubber average particle diameter = 11 mm. Fuel amount: 0.076 kg/s
corresponding to 2% of the total energy input to the kiln system.
165
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
154
The long conversion time for the tire char reduces the energy input in the material inlet end to no
more than 2% if complete conversion of the tire char is required in the rotary kiln. This low
figure is not in correspondence with the industrial experience from some calciner kilns, where it
is possible to feed around 10% of the total energy into the material inlet end of the rotary kiln. It
may indicate that the model under-estimates mass transfer of oxygen in the rotary kiln. Other
reasons for the deviation between model predictions and industrial experience may be that some
of the char is entrained in the flue gas and transported back towards the material inlet end of the
rotary kiln. Or it may be that a fraction of the char leaves the rotary kiln with the clinker and is
fully oxidized in the clinker cooler where the clinker is cooled with ambient air. The material
residence time in the clinker cooler is up to 30 min and the oxygen concentration is 21 vol.-%,
which provide good conditions for oxidation of unburned chars from the rotary kiln. However,
the material temperature will decrease from around 1450C to 100C through the clinker cooler,
which may entail considerable reduction of the char oxidation rate.
The model may be used to estimate the required kiln length in order to reach full conversion for
different fuel loads, and under different conditions. Figure 7-7 is an example where the required
kiln length is showed versus the energy percentage of the total required energy necessary to
produce cement. The fuel is tire-derived fuel, TDF. Values used in the model are summarized in
Appendix B and G. Four different excess air ratios from the main burner are shown, and the
corresponding oxygen concentrations in the flue gas from the main burner are shown in
parenthesis. At the lowest excess air ratio of 1.14, it is observed that 94 m and 142 m of kiln
length is required to reach full conversion of 1% and 2.5% TDF, respectively. Thus at low excess
air ratios, it is only possible to utilize small amounts of TDF in the material inlet end. But if the
bulk oxygen concentration increases, the energy share in the material inlet end can be increased
significantly. For the highest studied excess air ratio of 1.95, the energy share by TDF may be up
to 26% in a 150 m long rotary kiln or around 7% in a 50 m long rotary kiln.
166
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
155
0 25 50 75 100 125 150
Required rotary kiln length [m]
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
E
n
e
r
g
y
s
h
a
r
e
b
y
T
D
F
[
%
]
= 1.14
(y
O2
= 2.5 vol.-%)
= 1.33
(y
O2
= 5 vol.-%)
= 1.58
(y
O2
= 7.5 vol.-%)
= 1.95
(y
O2
= 10 vol.-%)
Figure 7-7: Model predicted required rotary kiln lengths to reach full conversion of TDF at different energy loads in
the material inlet end of rotary kilns. The fuel is TDF with average particle diameter of 11 mm. Different oxygen
concentrations are specified in the figure.
7.2.2 Conclusions for modeling solid fuel combustion in industrial rotary kilns
The fuel conversion model has been applied on an industrial rotary kiln to predict the conversion
time for tire rubber. The results may be used to estimate how far different fuel particles will be
transported into the rotary kiln as a function of particle size and key process parameters.
From the models results it is clear that the fuel devolatilization is a relatively fast process, taking
place in the first part of the material inlet end of the rotary kiln. By contrast, the char conversion
time is significantly longer and the char will thus be transported deep into the rotary kiln before
reaching full conversion. The reason for the slow char conversion is mass transfer limitations of
oxygen to the char particles.
The models may be used to estimate how large a share of the total energy input to the kiln
system that can be utilized into the material inlet end by considering e.g. the excess air ratio and
conversion time. However, when comparing full-scale experience with model predicted possible
energy shares in the material inlet, it seems that the model under-estimates the mass transfer of
oxygen to the char and thus, under-estimates how large fuel amounts it is possible to utilize in
the material inlet end.
Ideally, the models should be validated against experimental data from a cement plant. It is,
however, a complicated task to determine the conversion times of solid fuels fired into the
material inlet of a cement rotary kiln due to the extreme environment and poor accessibility. It is
also of interest to further develop the model regarding heat transfer between gas, wall and bed
along the kiln length in order to obtain more realistic predictions of temperature profiles.
167
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
156
7.3 Suggestions for alternative fuel utilization
It was shown in section 7.2 that the fuel devolatilization is a fast process while the char oxidation
is slow due to mass transfer limitations in rotary kilns. This was also confirmed by the
experimental results with tire rubber and pine wood combustion described in chapter 5. In
addition, the results from chapter 6 clearly showed that sulfur release from raw materials during
fuel combustion primarily took place during fuel devolatilization. It was also shown that the
sulfur release was most significant for several small fuel particles relative to few large fuel
particles due to a faster devolatilization, a result that corresponds well with industrial experience.
From these results the following recommendations regarding fuel utilization in the material inlet
end are:
1) The fuel should have a high volatile content.
2) Few large fuel particles are preferred rather than several small fuel particles. However,
the fuel particle size may be limited in order to keep the conversion time shorter than the
residence time in the rotary kiln.
3) The fuel and cement raw materials should be in minimum physical contact during fuel
devolatilization and/or the raw material temperature should be as low as possible during
fuel devolatilization.
4) Chars should preferably be of the crackling type in order to produce small char particles
that require shorter residence times than larger char particles. This corresponds to
pathway f in Figure 5-3 in chapter 5.
In the following, different relevant fuel types are considered with respect to the above mentioned
criteria. Please refer to table 3-2 in chapter 3 for details about each fuel type.
Tire-derived fuels (TDF)
TDF is a suitable fuel: It contains around 54-66 wt.-%volatiles and the char rapidly cracks into
smaller char particles, leading to an overall short conversion time. In addition, a high heating
value of 31-33 MJ/kg means that the required fuel mass is quite low. TDF contains around 1.5
wt.-%sulfur and will therefore increase the sulfur input to the kiln system, which may affect
process stability. As shown in section 7.1, TDF is an often used alternative fuel in the material
inlet end of cement rotary kilns.
Plastics
Plastics, except PVC, are excellent fuels since their volatile content approaches 100 wt.-%
leading to very short conversion times. Their heating values are also typically 40-44 MJ/kg,
which is very high and attractive.
Wood/biomass
Wood and different types of biomass have high volatile contents around 75-83 wt.-%. But their
char and ash matrix structures are often quite stable, leading to conversion according to the
shrinking particle or unreacted shrinking core mechanism, pathways G and H in Figure 5-3 in
168
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
157
chapter 5. This may lead to quite long char conversion times thereby limiting the fuel amounts to
be utilized in rotary kilns. In addition, the heating values are relatively low, around 18-19 MJ/kg,
and the moisture content may be very high. Pre-treatment in the form of drying or shredding may
be required in order to utilize large amounts of wood and biomass.
Meat and bone meal (MBM)
Meat and bone meal has a high volatile content around 65-72 wt.-%. The fixed carbon content is
typically low, in the order of 7-10 wt.-% which may make MBM an advantageous fuel.
However, the ash content is high, around 17-28 wt.-%, and the ash contains relatively high
amounts of sulfur, chlorine and phosphorous which may affect the process stability and product
quality. In addition, the heating value is relatively low, around 16-20 MJ/kg.
Refuse derived fuel (RDF)
RDF is a mixture of different materials and may exhibit great variations in volatile, char, ash,
moisture and chlorine content. Heating values for RDF are often also low, around 14 MJ/kg. The
utilization of RDF in rotary kilns should therefore be based on a worst case scenario to allow
flexibility for fluctuations in composition and amounts.
Sewage sludge (SS)
Sewage sludge has typically a high volatile content of 48-64 wt.-% and a low fixed carbon
content of 0-6 wt.-%. This has the consequence that sewage sludge burns fast in the rotary kiln
material inlet end, leaving only small amounts of char to be transported into the rotary kiln.
Sewage sludge may be characterized as a suitable fuel in the material inlet. However, the high
ash content of 17-50 wt.-% may affect the clinker quality in the form of e.g. phosphorous or
coarse silicate grains from the sewage sludge being mixed into the clinker, see section 3.4.1. The
lower heating value is typically 14-19 MJ/kg which is relatively low.
7.4 Conclusions for chapter 7
Industrial experience with alternative fuel utilization in the material inlet end of cement rotary
kilns indicates that the energy share in this point may be around 10% in modern ILC calciner
kilns, without jeopardizing process stability. A higher energy share may be achieved in long wet
and dry kilns or in SP kilns due to longer residence times in the rotary kiln and lower
temperatures in the material inlet end compared to ILC kilns.
A model has been developed to predict fuel combustion times for solid alternative fuels. The
model predicts quite long conversion times for TDF which will limit the amounts of fuels to be
used in the material inlet end. By comparison with industrial experience it is likely that the
model under-estimates mass transfer of oxygen to fuel chars in industrial rotary kilns, thereby
predicting too long conversion times.
169
Chapter 7 Analysis of industrial experience and combustion model for industrial rotary kilns
158
References
Bhatty, J. I., Miller, F. McGregor and Kosmatka, S. H.; Innovations in Portland Cement
Manufacturing, Portland Cement Association, ISBN: 0-89312-234-3, 2004.
Green-Andersen, P.; Kiln Performance Study, Internal Report, FLSmidth, Valby, Denmark, June
2008.
Hess, K., Omar, H., Rivera. L., Stubenbort, B. and Stutsman, B.; Pyroprocessing Audit,
FLSmidth, Bethlehem PA, USA, October 2008.
Jannerup, H.-E.; Personal communication, FLSmidth, Valby, Denmark, September 2009.
Nielsen, K. P. and Jns, E.; Internal report, FLSmidth, Valby, Denmark, September 2000.
Jns, E.; Behaviour of volatile matter, Lecture 05-02, International Cement Production Seminar,
FLSmidth, Denmark, 2007.
Larsen, M. B.; Alternative Fuels in Cement Production; PhD Thesis, Department of Chemical
Engineering, Technical University of Denmark, 2007. ISBN 978-87-91435-49-8.
Lewis, B. and Shenk, R.; Internal report, FLSmidth, Bethlehem PA, USA, June 2009.
Lund, J. E.; Direct communication. Plant Director, Cemex Philippines, Tinaan, Naga, Cebu,
Philippines, 2010.
Paone, P.; Personal communiation, FLSmidth, USA, September 2009.
Rosholm, P.; Personal communication, FLSmidth, Valby, Denmark, September 2009.
Snderborg, H. R.; The International Cement Production Seminar 2007, Volume IA, Lecture 05-
01; FLSmidth A/S, Valby, Denmark, 2007.
Turns, S. R.; An introduction to combustion. McGraw-Hill Int. Ed.2
nd
ed., 2006. ISBN: 978-007-
126072-5.
Valchev, P.; Personal communication, HOLCIM, Beli Izvor, Bulgaria, July 2010.
170
Chapter 8 Final conclusions
159
8 Final conclusions
An improved understanding of utilization of solid alternative fuels in the material inlet end of
cement rotary kilns has been gained through this PhD project. This has been obtained through
literature studies, experimental investigations, mathematical modeling, analysis of existing
process equipment and industrial experience.
8.1 Concluding trends
The main conclusions to be drawn from the literature review are:
1) Cement production is a very energy-intensive industry, accounting for approximately 2%
of the worlds primary energy consumption and being responsible for around 5% of the
worlds total CO
2
emissions.
2) The cement industry has a great interest in replacing fossil fuels with alternative fuels in
order to a) reduce overall production costs and b) minimize CO
2
emissions.
3) Solid alternative fuels utilization requires new solutions in the form of combustion
equipment or modifications of existing plant layout due to the fact that common
alternative fuels have different physical and chemical properties than traditional solid
fossil fuels.
4) The use of alternative fuels may influence clinker quality and/or process stability: The
clinker quality may be affected by minor components from the fuel ashes or from
unburned carbon leaving the rotary kiln with the clinker. The process stability may be
affected by increased circulation of inorganic volatiles, particularly sulfur and chlorine, in
the kiln system. Sulfur and chlorine containing species are known to cause deposit build-
ups in the kiln system, restricting gas and material flows.
The main trends regarding alternative fuel utilization in the material inlet end of cement
rotary kilns are:
1) Solid alternative fuel particles are rapidly mixed into cement raw materials. Thus, heat
transfer by conduction from the cement raw materials to the fuel particles must be
expected to be a major heat transfer mechanism rather than convection or radiation from
the freeboard gas above the material bed. Consequently, the temperature of the cement
raw materials becomes of great importance for the heating of the fuel particles. In
addition, it must be expected that mass transfer of oxygen from the freeboard gas to the
fuel particles will be hindered by the cement raw materials covering the fuel particles.
The mixing process depends on several parameters such as fuel particle size and shape,
fuel particle density, bed fill degree and kiln rotational speed, making the mixing process
complicated to model.
2) Heating, drying and devolatilization of solid alternative fuels are fast processes which
primarily depend on temperature and fuel particle size. The volatile content of alternative
fuels is typically high. Consequently, a large fraction of the energy from the alternative
fuels will be released rapidly in the material inlet end of the rotary kiln. This creates a
171
Chapter 8 Final conclusions
160
risk for increasing local temperatures and reducing conditions in the raw material charge,
which consequently may lead to increased sulfur release from the raw materials, followed
by increased formation of deposit build-ups in the kiln system. The sulfur release may be
kept low during fuel devolatilization by minimizing the physical contact between fuel and
raw materials, increasing fuel particle size, increasing the bulk oxygen concentration or
by lowering the raw material temperature in the material inlet end of the rotary kiln.
3) Char oxidation is a slow process and therefore has a great influence on the amounts of
fuel to be utilized, due to limited residence time in cement rotary kilns. Several
parameters control the rate of char oxidation: a) bulk oxygen concentration, b) mass
transfer rate of oxygen to char particles, c) conversion pathway, d) bed material fill
degree and e) char particle size and shape. Parameters such as temperature and rotational
speed only have a minor influence on char oxidation under the conditions in cement
rotary kilns.
4) When considering alternative fuel utilization in the material inlet end of cement rotary
kilns, the key parameters are a) overall input of the inorganic volatiles Cl, S, K and Na to
the kiln system, b) volatile and char content of the fuel, c) (char) conversion pathway and
d) fuel particle size and shape. Rotary kiln specific parameters such as length, inner
diameter, bed fill degree, solids residence time and gas velocity has great influence on the
amounts of fuel it is possible to utilize.
8.2 Suggestions for further work
Much effort has been spent on the development of a new, high-temperature experimental set-up,
capable of simulating the conditions in the material inlet end of a cement rotary kiln. This work
has been at the expense of the time available to make experiments in the rotary drum. The rotary
drum is a unique experimental set-up which enables us to study several interesting parameters,
such as combustion behavior of solid fuels and inorganic chemistry under well-defined
conditions. It is therefore strongly recommended to further explore the many opportunities and
continue the experimental work in the rotary drum set-up.
The char combustion model presented in this work has not been optimized with respect to
predicting the effect of bed fill degree and temperature on the conversion time. A comparison
between the model and full-scale experience also seems to imply that the model over-estimates
the time for external mass transfer of oxygen to the char. Hence it will be necessary to further
optimize the char combustion model. Ideally, the model should be validated against full-scale
data.
Results from the fuel/raw material mixing experiments indicated that insertion of mixing devices
in rotary kilns could help to optimize fuel-gas contact, thereby enhancing mass transfer of
oxygen to fuel chars. This opportunity should be further investigated, e.g. by experiments in the
rotary drum set-up or by full-scale tests.
172
List of symbols
161
List of symbols
Roman nomenclature
A Area m
2
Bi Biot number -
C
0
Moisture content g water/g dry fuel
C
p
Specific heat capacity J/(kg K)
C
2
S Belite -
C
3
S Alite -
C
4
AF Ferrite -
C
3
A Aluminate -
D Diameter of drum/kiln m
D Diffusion coefficient m
2
/s
d Particle diameter m
E
a
Activation energy J/mol
F Bed fill degree - (%)
FC Fixed carbon wt.-%
g Gravitational speed m/s
2
h Heat transfer coefficient W/(m
2
K)
h Height m
ILC In-Line calciner -
k
p
Thermal conductivity W/(m K)
k
g
Mass transfer coefficient m/s
k
0
Rate constant s
-1
L Length m
LHV Lower heating value MJ/kg
m Mass g (kg)
M Molar weight mol/kg
MBM Meat and bone meal -
n Rotational speed min
-1
n Moles Mol
R Universal gas constant 8.314 J/(mol K)
R Radius of drum/kiln m
r Particle radius m
r Rate mol O
2
/ (m
2
s)
RDF Refuse derived fuel -
Re Reynolds number -
P Pressure Pa
P Probability %
PE Poly ethylene -
PVC Poly vinyl chloride -
Sc Schmidt number -
Sh Sherwood number -
SLC Separate Line calciner -
SP Suspension preheater -
t Time s or min
T Temperature K or C
TDF Tire-derived fuel -
V Degree of devolatilization -
173
List of symbols
162
V Volume m
3
VM Volatile matter wt.-%
Y Distribution factor in bed -
y Mole fraction -
X Degree of conversion -
x Distance m
Greek nomenclature
Thermal diffusitivity m
2
/s
Mol C in fuel Mol C/kg fuel
Mol H in fuel Mol H/kg fuel

i
Positive solutions to equation -
Porosity -
Emissitivity -
Density kg/m
3
or mol/m
3
Mol N in fuel Mol N/kg fuel
Mol O in fuel Mol O/kg fuel
Sphericity -
Excess air ratio -
Heat of vaporization Cal/g
Mol S in fuel Mol S/kg fuel
Rotational speed rad/s
Stefan-Boltzmann constant 5.6704 10
-8
W/(m
2
K
4
)
CO/CO
2
product ratio -
Mol Cl in fuel Mol Cl/kg fuel
Total time s
Tortuosity -
Mol H
2
O in fuel Mol H
2
O/kg fuel
Subscripts
a Average
b Bed
c Char
e Wet/dry interface
Cond Conduction
Conv Convection
Devol Devolatilization
eff Effective
g Gas
p Particle
Rad Radiation
Sph Spherical
Vol Volatiles
w Wall
0 Initial value
174
Appendixes
163
Appendixes
175
Appendixes
164
Appendix A Sensitivity analysis on estimated fuel conversion
Several random errors may affect the predicted degree of fuel conversion. The impact of these
accumulated random errors is analyzed with the accumulation law of uncertainties (Larsen and
Hellesen, 1998). The degree of fuel conversion is estimated from the ideal gas law:
2
0
,
( )
t
carbon
Fuel CO CO
carbon total
M P V
X t y y dt
m R T

= +

}
(A-1)
When calculating the degree of fuel conversion a series of calculated and measured values is used
and the fuel conversion may be expressed as a function of these values:
( )
2
, , ,
Fuel Fuel CO CO
X X T V y y = (A-2)
The uncertainty on the determined fuel conversion, (X
Fuel
), is then predicted with the accumulation
law of uncertainties:
( ) ( ) ( ) ( ) ( )
( ) ( )
( )
( )
2
2
2 2
2
2
2 2
2 2
2
0 0
,
0
Fuel Fuel Fuel Fuel
Fuel CO CO
CO CO
t t
CO CO CO CO
carbon
t
carbon total
CO CO CO
X X X X
X T V y y
T V y y
PV P
y y dt T y y dt V
RT RT M
m
PV
y y dt y
RT
o o o o o
o o
o
| |
| | c c c c | | | |
= + + + = |
| | |
|
c c c c
\ . \ .
\ .
\ .
| | | |
+ + + +
| |
\ . \ .

|
+

\
} }
} ( )
( )
2 2
2 2
0
t
CO CO CO
PV
y y dt y
RT
o
| | |
+ +
| |
. \ .
}
(A-3)
In the estimation of the overall uncertainty, the temperature is assumed to fluctuate 1%, the gas
flow is assumed to fluctuate 3% and the uncertainties on measured gas species is assumed to
fluctuate 1% for each species. The pressure P is assumed to be constant.
The total uncertainty on the fuel conversion has been estimated for typical experiments with
combustion of tire rubber and pine wood. The uncertainty for one of these experiments with pine
wood is presented at different fuel conversion degrees in Table A-1. It is observed that the total
uncertainty increases with time, and reaches 0.059 at full fuel conversion, corresponding to a
relative uncertainty of 5.9%. The measured gas flow has the greatest impact on overall
uncertainty. It accounts for 87% of the total uncertainty.
X
Fuel
0.25 0.5 0.75 1.0
(X
Fuel
) 0.033 0.043 0.051 0.059
Table A-1: Total uncertainty of fuel conversion at different conversion degrees.
176
Appendixes
165
Appendix A references:
Larsen, M. O. and Hellesen, B.; Statistik 1, Institut for Anvendt Kemi, 3rd ed., 1998. Technical
University of Denmark. (in Danish).
177
Appendixes
166
178
Appendixes
167
Appendix B Data used in the combustion models
Property TDF Pine wood References
Specific heat capacity, C
p
, [J/(kg
K)]
2,000 1,500 Larsen et al., 2006
Leon et al., 2000
Initial particle density,
p
, [kg/m
3
] 1,150 690 Yang et al., 1995
Leon et al., 2000
k
p
, virgin fuel thermal conductivity,
[W/(m K)]
0.30 0.2 Sellassie et al.,
2007
Leon et al., 2000
Conductive heat transfer coefficient,
h
CBP
, [W/(m
2
K)]
250 250 Linjewile et al.,
1993
Convective heat transfer coefficient,
h
CBP
, [W/(m
2
K)]
150 150 Tscheng and
Watkinson, 1979
Table B-1: Fuel specific data used in the devolatilization models.
Fuel TDF Pine wood References
Property

p
[kg/m
3
] 1,100 690 Larsen et al.,
2006
Leon et al., 2000
k
0
[1/s] 1.5 ( 5.1) 10
10
1.19 10
8
Larsen et al.,
2007
Kastanaki and
Vamvuka, 2006
Ea [kJ/mol] 19328 14010 Larsen et al.,
2007
Kastanaki and
Vamvuka, 2006
Exponent m 0.63 0.89 Larsen et al.,
2007
Kastanaki and
Vamvuka, 2006
LHV [MJ/kg] 37 14 From analysis
Table B-2: Fuel specific data used in the char oxidation models.
Appendix B references
Kastanaki, E. and Vamvuka, D.; A comparative reactivity and kinetic study on the combustion of
coal-biomass char blends. Fuel, Vol. 85, 1186-1193, 2006.
Larsen, M. B., Schultz, L., Glarborg, P., Skaarup-Jensen, L., Dam-Johansen, K., Frandsen, F. and
Henriksen, U.; Devolatilization characteristics of large particles of tyre rubber under combustion
conditions. Fuel, vol. 85, 1335-1345, 2006.
179
Appendixes
168
Larsen, M. B., Hansen, M. L., Glarborg, P., Skaarup-Jensen, L., Dam-Johansen, K. and Frandsen,
F.; Kinetics of tyre char oxidation under combustion conditions. Fuel, vol. 86, 2343-2350, 2007.
Leon, G., Cruz-de-Leon, J. and Villasenor, L.; Thermal characterization of pine wood by
photoacoustic and photothermal techniques. Holz als Roh- und Werkstoff, 58, 241-246, 2000.
Linjewile, T. M., Hull, A. S. and Agarwal, P. K.; Heat transfer to a large mobile particle in gas-
fluidized beds of smaller particles. Chem. Eng. Sci., 48, 21, 3671-3675, 1993.
Sellassie, K. G., Moo-Young, H. K. and Lioyd, T. B.; Determination of the thermal conductivity of
shredded tyres by utilizing a hot plate apparatus. Int. J. Environment and Waste Management, Vol.
1, Nos. 2/3, 179-191, 2007.
Tscheng, S. H. and Watkinson, A. P.; Convective heat transfer in a rotary kiln. Can. J. Chem. Eng.,
Vol. 57, 433-443, August 1979.
Yang, J., Tanguy, P. A. and Roy, C.; Heat transfer, mass transfer and kinetics study of the vacuum
pyrolysis of a large used tire particle. Chem. Eng. Sci., Vol. 50, No. 12, 1909-1922, 1995.
180
Appendixes
169
Appendix C Calculation of values used in the char model
The char particle density,
p
, is calculated based on knowledge of the virgin fuel density as well
as volatile and ash content:
( )
,0
1
p p
Volatiles Ash = (C-1)
The gas phase O
2
-concentration, C
O2
, is assumed to be constant in the rotary kiln freeboard cross
section due to turbulent flow conditions, and is found from the ideal gas law:
2
2
O
O
y P
C
R T

(C-2)
where P is the pressure in Pa, R is the gas constant and T is the temperature in Kelvin. y
O2
is the
mole fraction of O
2
in the gas phase.
The binary diffusion coefficient, D
O2
, for oxygen in nitrogen is temperature dependent. The
value at a specific temperature can be predicted with an accuracy of 5% by use of the Chapman-
Enskog kinetic theory (Bird et al., 2002):
2
2 2
3
2
1 1 1
0.0018583
O
O N
D T
M M Po
| |
= + |
|
O
\ .
(C-3)
where T is the temperature in Kelvin, M is the molar masses of oxygen and nitrogen,
respectively, P is the pressure in atm, is the collision diameter in ngstrm and is the
collision integral for diffusion. The unit of D
O2
will be cm
2
/s.
By using table values from Bird et al., the so-called Lennard-Jones parameters, and /, can be
calculated (Bird et al., 2002):
( ) ( )
2 2
10
3.433 3.667 3.55 10
O N
m o o o

= + = + = (C-4)
2 2
113 99.8 106.2
O N
K
c c
c
k k k
= = = (C-5)
The collision integral can be calculated Irom the expression:
*0.1561 * * *
1.06036 0.193 1.03587 1.76474
exp(0.47635 ) exp(1.52996 ) exp(3.89411 ) T T T T
O = + + + (C-6)
where
*
T
T
k
c
= (C-7)
181
Appendixes
170
By using these expressions, D
O2
can be calculated at any given temperature.
Appendix C references
Bird, R. B., Stewart, W. E. and Lightfoot, E. N.; Transport Phenomena, 2
nd
edition, John Wiley
& Sons, 2002. ISBN: 0-471-41077-2.
182
Appendixes
171
Appendix D Species included in the equilibrium calculations
For the equilibrium calculations, gas phase compounds, pure solid phases, and pure liquid phases
were included from the FactSage 6.0 compound database. The species included from the
compound database are listed below.
Gas phase components
N
2
N
2
O H OAlOH
CO
2
HONO HNO OAlCl
H
2
O ONCl K CaCl
O
2
CO O
2
S(OH)
2
(FeCl)
2
HCl SO
3
NaO NaH
KCl K
2
SO
4
NO
3
N
(KCl)
2
O (NaOH)
2
SO
2
Cl
2
(NaCl)
2
Na
2
SO
4
(KOH)
2
FeO
NO H
2
KO CaOH
Cl HOOH Cl
2
O HNCO
NO
2
FeCl
2
KAlCl
4
NH
2
Cl
2
NaFeCl
4
NaAlCl
4
KH
OH HONO
2
AlCl
3
SOCl
2
HOCl FeCl
3
N
2
O
3
NH
CaCl
2
Ca(OH)
2
COCl N
2
O
5
KOH Na COCl
2
FeCl
NaOH Fe(OH)
2
HCOOH (FeCl
3
)
2
ClO O
3
NH
3
AlCl
2
SO
2
NO
2
Cl SO N
3
HOO ClO
2
N
2
O
4
HCO
CaO SCl NS SiHCl
3
H
2
CO HCN NaCN SiCl
2
Fe S SiCl
3
CH
2
Cl
2
Na
2
AlO
2
KCN N
2
H
4
NCO Ca AlO CH
3
OH
SiCl
4
AlOH Al
2
Cl
6
CaH
HNNH AlCl CCl
4
SSO
SCl
2
HS CHCl
3
CH
3
Cl
K
2
H
2
S CCl
3
AlOH
ClCN COS CN CH
4
AlCl
6
Fe SiO CCl
2
CNN
S
2
Cl C
2
HCl CS
2
CH
3
CH
2
OH
C
2
O Si CH
3
NC C
2
H
5
Cl
CS CH
3
NH
2
CH
2
CHCl CH
3
N
2
H
3
CaS Al
2
O (KCN)
2
C
2
H
6
(AlO)
2
SiH
3
Cl C
2
H
4
O AlC
SiCl C
2
Cl
4
CH
3
SH AlN
CH C
2
H
4
C
2
H
4
Ca
2
ClSSCl CHClCCl
2
C
2
Cl
6
(NaCN)
2
183
Appendixes
172
CH
3
COOH SiH C
2
Cl
5
H C
2
C
3
O
2
C
2
H
4
O S
3
CH
3
CHCl
2
HCNS SiS SiH
4
SiN
(CN)
2
AlS C
2
H
3
CH
3
COCl
C
2
Cl
2
C
2
N CHCl
2
CHCl
2
C
2
H
H
2
S
2
CH
2
CCl
2
CHCl2CH
2
Cl SiCH
3
Cl
3
Pure solid phases
Ca
2
SiO
4
NaOH NaNO
2
Fe
CaSO
4
FeOCl Fe
2
O
3
(H
2
O) KAlSi
3
O
8
Ca
2
Al
2
SiO
7
Ca
3
Al
2
Si
3
O
12
KNO
3
NaAlSi
3
O
8
Ca
2
Fe
2
O
5
SiO
2
NaHCO
3
K
CaAl
2
O
4
Ca
3
Fe
2
Si
3
O
12
CaOCl
2
Al
6
Si
2
O
13
KCaCl
3
CaAl
12
O
19
FeSiO
3
(FeO)
2
(SiO
2
)
NaCl KAl
9
O
14
FeCl
2
K
2
(OH)(NO
3
)
KCl Fe
2
O
3
Na
2
Al
12
O
19
(FeO)
2
(SiO
2
)
Ca
2
SiO
4
KOH FeCO
3
NaAlSi
3
O
8
CaAl
4
O
7
(CaSO
4
)
2
(H
2
O) FeCl
2
FeSiO
3
CaFe
2
O
4
(Na
2
O)(SiO
2
) CaAl
2
Si
2
O
8
FeCl
3
CaO Na
3
(OH)(SO
4
) FeSiO
3
K
2
O
2
CaCO
3
K
2
Ca(CO
3
)
2
KFeCl
4
KAl(SO
4
)
2
Ca
3
SiO
5
NaAl
9
O
14
KHCO
3
H
2
Si
2
O
5
Ca
3
Al
2
O
6
K
2
CO
3
NaAlSi
2
O
6
KAlCl
4
CaCl
2
CaAl
2
Si
2
O
8
H
2
SiO
3
NaAlSi
2
O
6
H
2
O
Ca
3
Si
2
O
7
NaAlSiO
4
Al(OH)
3
H
4
SiO
4
Al
2
O
3
Na
2
Ca
2
Si
3
O
9
(Na
2
O)(SiO
2
)
2
Na
2
Ca
3
Si
6
O
16
Na
2
SO
4
Al
2
SiO
5
Fe(OH)
2
SO
3
CaSiO
3
H
2
O CaFeSi
2
O
6
K
2
O
Fe
2
O
3
Al
2
SiO
5
K
2
Si
2
O
5
Na
6
Si
2
O
7
CaSiO
3
K
2
Ca
2
(CO
3
)
3
FeSO
4
Na
3
(OH)
2
(NO
3
)
K
2
SO
4
Ca
2
FeSi
2
O
7
K
2
Si
2
O
5
Ca(NO
3
)
2
K
2
Ca
2
(SO
4
)
3
(CaSO
4
)
2
(H
2
O) Ca
2
Al
3
Si
3
O
12
(OH) NaAlCl
4
KAlO
2
CaSO
3
CaSO
4
(H
2
O)
2
NaH
CaFe
4
O
7
OAlCl Na
2
SO
3
KAl
3
Si
3
O
10
(OH)
2
CaO
2
Al
2
O
3
(H
2
O) Na
4
CaSi
3
O
9
C
CaAl
2
SiO
6
Al
2
O
3
(H
2
O) Fe
3
O
4
CaAl
2
Si
2
O
7
(OH)
2
(H
2
O)
NaAlO
2
KAlSi
2
O
6
(Na
2
O)
2
(SiO
2
) KClO
4
K
3
Na(SO
4
)
2
K
2
SiO
3
Fe(OH)
3
KH
Al
2
Fe
2
O
6
K
2
SiO
3
NaFeCl
4
NaAl
3
Si
3
O
12
H
2
Ca(OH)
2
K
2
FeCl
4
Na
2
O AlCl
3
NaAlSiO
4
NaO
2
Na
2
O
2
KClO
4
KAlSiO
4
K
2
Al
12
O
19
AlCH
2
NaO
5
CaSO
3
(H
2
O)
2
Na
2
CO
3
KFeCl
3
K
2
SO
3
Al
2
O
3
(H
2
O)
3
Na
2
CO
3
KO
2
Na
2
O
2
K
2
Si
4
O
9
KAlSiO
4
NaNO
2
Na (CaO)
6
(SiO
2
)
6
(H
2
O)
FeO KNO
3
(FeO)
2
(SiO
2
) Na
2
CaSi
5
O
12
184
Appendixes
173
Fe
3
O
4
NaNO
3
Na
2
OHNO
3
K
2
Si
4
O
9
NaOH (Na
2
O)(Fe
2
O
3
) CaAl
4
Si
2
O
10
(OH)
2
NaClO
4
FeAl
2
O
4
NaNO
3
KAlSi
3
O
8
(CaO)(SiO
2
)
2
(H
2
O)
2
S CaC
2
O
4
(H
2
O) (Al
2
O
3
)(SiO
2
)
2
Ca
AlN N
2
O
4
H
6
Si
2
O
7
(CaO)
8
(SiO
2
)
6
(H
2
O)
3
(CaO)
3
(SiO
2
)
2
(H
2
O)
3
(Al
2
O
3
)(SiO
2
)
2
(H
2
O)
2
CaS (Al
2
O
3
)(SiO
2
)
2
(H
2
O)
2
NH
4
Cl KFe
3
AlSi
3
O
10
(OH)
2
CaCl
2
(H
2
O)
6
(CaO)
5
(SiO
2
)
6
(H
2
O)
3
(Al
2
O
3
)(SiO
2
)
2
(H
2
O)
2
Fe
2
N K
4
C
2
O
6
(H
2
O)
3
NH
4
NO
3
Fe
3
Al
2
Si
3
O
12
FeS Al Ca(NO
3
)
2
(H
2
O)
3
Al
2
Si
4
O
10
(OH)
2
Fe
2
(SO
4
)
3
K
2
S NH
4
HCO
3
N
2
O
5
Al
2
(SO
4
)
3
(CaO)
3
(Al
2
O
3
)(H
2
O)
6
Si
NaCN Ca(NO
3
)
2
(H
2
O)
2
CaH
2
Na
6
Si
8
O
19
KCN Na
2
S Fe
2
Al
4
Si
5
O
18
Ca(NO
3
)
2
(H
2
O)
4
AlS FeSi FeAl
2
Cl
8
FeS
2
NH
4
Al(SO
4
)
2
NH
4
ClO
4
Na
2
S
2
SiS
(CaO)
12
(SiO
2
)
6
(H
2
O)
7
CO(NH
2
)
2
AlCl
3
(H
2
O)
6
FeSO
4
(H
2
O)
7
NH
4
ClO
4
NaS
2
(CaO)
4
(SiO
2
)
6
(H
2
O)
5
SiC
Fe
4
N AlH
3
Fe
3
C CaC
2
(NH
4
)
2
(SO
4
) CaSi NH
4
HS (CaO)
4
(Al
2
O
3
)(H
2
O)
13
C
2
Na
2
Na
2
SO
4
(H
2
O)
10
Na
2
CO
3
(H
2
O)
10
Al
2
(SO
4
)
3
(H
2
O)
6
CH
3
NH
3
Cl Ca
3
N
2
Ca(ClO
4
)
2
(H
2
O)
4
NH
2
CH
2
COOH
SiS
2
(CaO)
2
(Al
2
O
3
)
2
(SiO
2
)
8
Fe
3
Si Na
2
S
3
Pure liquid phases
KCl Na
2
SO
4
H
2
O KOH
NaCl K
2
SO
4
Al
2
O
3
SiO
2
CaSO
4
CaO NaOH K
2
CO
3
CaCl
2
NaAlO
2
KAlO
2
(Na
2
O)(SiO
2
)
CaAl
2
O
4
CaAl
4
O
7
FeO Fe
3
O
4
NaNO
3
NaNO
2
K
2
SiO
3
K
2
Si
2
O
5
K
2
SO
3
KNO
3
(Na
2
O)(SiO
2
)
2
Na
2
SO
3
FeCl
3
Na Na
2
O (Na
2
O)
2
(SiO
2
)
HOOH KAlCl
4
NaAlCl
4
AlCl
3
K HONO
2
O
2
S(OH)
2
Na
6
Si
2
O
7
Fe SO
3
K
2
O H
2
SO
4
(H
2
O)
HCOOH N
2
O
4
H
2
SO
4
(H
2
O)
2
NH
4
Cl
K
2
Si
4
O
9
S Ca H
2
SO
4
(H
2
O)
3
NaCN H
2
SO
4
(H
2
O)
6
Al CH
2
Cl
2
KCN HCN K
2
S N
2
H
4
SiCl
4
Na
2
S CHCl
3
CH
3
OH
SCl
2
H
2
SO
4
(H
2
O)
4
CCl
4
AlS
FeS Na
2
S
2
SiHCl
3
(NH
4
)
2
O
CH
3
NO
2
ClSSCl H
2
S
2
CH
3
NH
2
NaS
2
CH
3
COOH CH
3
COCl C
2
Cl
4
CHClCCl
2
Fe
3
C CH
2
CCl
2
_cis-12 CH
2
CCl
2
_trans-12
CaC
2
C
2
H
4
O CH
2
CCl
2
C
2
Na
2
185
Appendixes
174
CS
2
CH
3
SH CHCl
2
CHCl
2
CH
3
CHCl
2
CH
3
NC C
2
H
4
O (CH
2
OH)
2
C
2
H
5
Cl
CH
3
CH
2
OH SiS
2
CH
3
N
2
H
3
Na
2
S
3
186
Appendixes
175
Appendix E Gas and bed temperatures through the rotary kiln
Gas and bed temperatures through the rotary kiln are assumed to change according to the
following equations:
( )
( ) ( )
, ,
,
1
fuel fuel g g out g in
Vol Vol Char Char
p g
fuel g fuel g
Vol Char
dT T T Y m y LHV m y LHV dV dX
dt C dt dt
m y V t m m y X t m
t
- -
- - - -
| |
|

| = +
| | | |
|
+ +
| | |
\ . \ . \ .
(E-1)
( ) ( )
, ,
,
fuel fuel
b out b in
b Vol Vol Char Char
p b
fuel b fuel b
Vol Char
T T
dT m y LHV m y LHV Y dV dX
dt C dt dt
m y V t m m y X t m
t
- -
- - - -
| |
|


| = + +
| | | |
|
+ +
| | |
\ . \ . \ .
(E-2)
These equations describe that the gas and bed temperatures through the rotary kiln changes
linearly, but are affected by the heat release from both the volatile and char content of the fuels
that are fed to the material inlet end. T
g,out
and T
b,out
are gas and bed temperatures in the material
outlet end of the rotary kiln, and T
g,in
and T
b,in
are gas and bed temperatures in the material inlet
end oI the rotary kiln. is the mean solid residence time in the rotary kiln, and is estimated by
the empirical formula:
11.2 L
n D s
t

=

(E-3)
where L is the rotary kiln length, n is the rotary kiln speed, D is the inner kiln diameter and s is
the rotary kiln inclination. This empirical formula can only be used to give a rough estimate of
the mean solid residence time in the rotary kiln, since it does not consider physical properties of
the solids. However, the mean residence time predicted by the formula is in good correspondence
with experimentally measured mean residence times in a cement rotary kiln reported by Strau et
al. (1987) and experiments with sand in a pilot-scale rotary kiln, reported by Liu and Specht
(2006).
Y is the fraction of heat that is released to the bed relative to the heat released to the gas, see
section 5.1 in chapter 5. V is the dimensionless devolatilization degree, and the change with time
is estimated from equation 5.6 or 5.9 in chapter 5:
1
Devol
dV
dt t
= (E-4)
X is the dimensionless char oxidation degree, and the change with time is estimated from
equation 5.27 in chapter 5. m
fuel
, m
g
and m
b
are the mass flows of fuel, gas and raw material bed
in kg/s. y
Vol
and y
Char
are the volatile and char fractions in the fuel, respectively. LHV
Vol
and
LHV
Char
are the lower heating values of volatiles and char, respectively. LHV
Char
is found from
187
Appendixes
176
the reaction enthalpy for oxidation of carbon to CO
2
, while LHV
Vol
is found by subtracting
LHV
Char
from the fuel lower heating value, LHV:
Char Char
Vol
Vol
LHV y LHV
LHV
y

= (E-5)
The first term in equations (E-1) and (E-2) refers to the situation in a rotary kiln without
alternative fuel utilization in the material inlet end, where it is assumed that bed and gas
temperatures change linearly along the kiln length. This is a simplification of actual conditions in
a rotary kiln where gas and bed temperatures will also be affected by gas-bed heat transfer, gas-
wall heat transfer and wall-bed heat transfer. These heat transfer mechanisms should be included
in the equations as well in order to obtain more realistic temperature profiles along the kiln
length. It is therefore important to point out that equation (E-1) and (E-2) should be further
developed to include the entire heat transfer mechanism between gas-bed, bed-wall, gas-wall and
heat loss to the surroundings. This work has not been made in the present study because char
oxidation is assumed to be the slowest and most critical part of the fuel combustion in industrial
rotary kilns, regarding conversion times: Temperature is expected to be of minor importance for
the fuel char oxidation which is more dependent on oxygen mass transfer to the char particles.
T
g, in
is calculated from the expression:
( )
, , ,0
,
1 fuel
g in g in
fuel g
p g
Y LHV m
T T
C m m
-
- -

= +
| |
+
|
\ .
(E-6)
where T
g, in,0
is the gas inlet temperature if there is no fuel utilization in the material inlet end.
Appendix E references
Liu, X. Y. and Specht, E.; Mean residence time and hold-up of solids in rotary kilns. Chem. Eng.
Sci., 61, 5176-5181, 2006.
Strau, F., Steinbi, E. and Wolter, A.; Measurement of retention times in cement burning
systems with the aid of radionuclides. ZKG-International, 9, 441-446, 1987.
188
Appendixes
177
Appendix F Stoichiometric calculations on flue gas composition
The following formula is used to describe the reaction of the fuel with oxygen from the
combustion air:
( )
2 2 2
2 2 2 2 2 2
79
( ) ( )
4 2 2 21
79
( ) ( ) 1
2 2 4 2 2 21
C H S N O Cl H O O N
CO H O SO NO Cl O N
o | o
| o
o
| | o
o o o
| |
+ + + + +
|
\ .
| |
+ + + + + + + + + +
|
\ .
(F-1)
where is the excess air ratio. The stoichiometric numbers , , etc., have the unit mol/kg fuel.
The reaction can be divided into one reaction equation for devolatilization and one for char
oxidation. The volatile fraction in the fuel is assumed to be all non-carbon compounds and the
amount of carbon needed to reach the volatile weight fraction, y
Vol
:
( )
, 2 2 2
2 2 2 2 2 2
79
( ) ( )
4 2 2 21
79
( ) ( ) 1
2 2 4 2 2 21
Vol Vol
Vol Vol
C H S N O Cl H O O N
CO H O SO NO Cl O N
o | o
| o
o
| | o
o o o
| |
+ + + + +
|
\ .
| |
+ + + + + + + + + +
|
\ .
(F-2)
where
Vol
is found from:
( )
2
Vol H S N O Cl H O
Vol
C
y M M M M M M
M
| o |
o
+ + + + +
= (F-3)
The reaction equation for the char oxidation becomes:
( )
, 2 2
2 2 2
79
21
79
1
21
Char Char
Char Char
C O N
CO O N
o
o
o o
| |
+ +
|
\ .
| |
+ +
|
\ .
(F-4)
where
Char
is found from:

Char
=
Vol
(F-5)
The minimum amount of air, N
air,min
, needed for the fuel combustion and the minimum amount
of produced flue gas, N
FG,min
, is estimated from:
2
,min
4 2 2
air air
O
N
y
| o |
o + + +
= (F-6)
189
Appendixes
178
,min ,min
4 2 2 2
FG air
N N
| o |
= + + + + + (F-7)
The excess air ratio is the actual amount of combustion air relative to the minimum amount:
,min
air
air
N
N
= (F-8)
The actual amount of produced flue gas is determined as:
( )
,min ,min
1
FG FG air
N N N = + (F-9)
The mole fraction of oxygen in the flue gas is estimated from:
( )
2
2
,min
1
air
air O FG
O
FG
N y
y
N

= (F-10)
The flue gas oxygen fraction in the kiln material inlet end may be described as:
2 2
, , burner
,
( )
4 2 2
fuel
FG FG
O in O
FG in
m
y y
n
| o
o
-
-
+ + +
= (F-11)
where y
O2,burner
is the oxygen fraction in the flue gas leaving the kiln main burner. n
FG,in
is the
total moles of flue gas per second.
The change in oxygen concentration along the kiln length can be estimated from the expression:
( )
( )
( )
( )
( ) ( )
2 2
,
,
,
( )
4 2 2
( )
4 2 2
( )
4 2 2
fuel
Vol
O O in
FG in fuel
fuel
Char
FG in fuel fuel
Char
m V t
y t y
n m V t
m X t
n m V t m X t
| o
o
| o

o
| o
o
-
- -
-
- - -
+ + +
= +
+ +

+
+ +
(F-12)
190
Appendixes
179
Appendix G Data used in the industrial model
Variable Industrial kiln
Kiln length, m 60
Inner refractory diameter, m 4.35
Speed of rotation, min
-1
3
Angle of tilt, 1.8
Solids flow, tpd (kg clinker/s) 3,500 (40.51)
Gas flow rate from burner, Nm
3
/kg clinker 0.42
Gas temperature, material inlet end, C 1,100
Gas temperature, material outlet end, C 1,600
Bed temperature, material inlet end, C 900
Bed temperature, material outlet end, C 1,450
Pressure, atm 1
Gas molar weight, g/mol 31.5
Excess air ratio at main burner, 1,33 (5 vol.-% O
2
in flue gas from main
burner)
Table F-1: Dimensions of kiln and operation conditions.
Parameter Value
Stefan-Boltzmann constant, , W/(m
2
K
4
) 5.6704 10
-8

W
0.9

p
0.9

B
, kg/m
3
1650
C
p,bed
, J/(kg-K) 800
Porosity, bed 0.5
h
COND
, bed-char particle, W/(m
2
K) 300
H
r
, C to CO, 25C, J/mol -110520
T
p,0
, surface temperature, initial char ox., C 1,050
Y (fuel contact ratio between gas and bed) 0.5
Table F-2: Values used in the calculations.
191
Appendixes
180
192
181
Paper A- Mixing large and small particles in a pilot-
scale rotary kiln
Published in Powder Technology, 210, 273-280, 2011.
Paper B High-temperature release of SO
2
from
calcined cement raw materials
Published in Energy & Fuels, 25, 2917-2926, 2011.
Paper C Sulfur release from cement raw materials
during solid fuel combustion
Published in Energy & Fuels, 25, 3917-3924, 2011.
Paper D Devolatilization and combustion of tire
rubber and pine wood in a pilot scale rotary drum
Published in Energy & Fuels, 26, 854-868, 2012.
193
Mixing large and small particles in a pilot scale rotary kiln
Anders Rooma Nielsen
a,b,
, Rasmus Wochnik Aniol
a,b
, Morten Boberg Larsen
a
,
Peter Glarborg
b
, Kim Dam-Johansen
b
a
FLSmidth A/S, Vigerslev All 77, DK-2500 Valby, Denmark
b
Department of Chemical Engineering, CHEC Research Centre, Technical University of Denmark (DTU), DK-2800 Lyngby, Denmark
a b s t r a c t a r t i c l e i n f o
Article history:
Received 5 October 2010
Received in revised form 23 November 2010
Accepted 27 March 2011
Available online 5 April 2011
Keywords:
Rotary kiln
Mixing
Alternative fuels
Cement
Heating
Combustion
The mixing of solid alternative fuel particles in cement raw materials was studied experimentally by visual
observation in a pilot scale rotary kiln. Fuel particles were placed on top of the raw material bed prior to the
experiment. The percentage of particles visible above the bed as a function of time was evaluated with the bed
predominantly in the rolling bed mode. Experiments were conducted to investigate the effects of fuel particle
size and shape, fuel particle density, rotary kiln ll degree and rotational speed. Large fuel particles and low-
density fuel particles appeared more on top of the bed than smaller particles and high-density fuel particles.
Fuel particle dimensions and sphericity were important parameters for the percentage of visible particles.
Increasing bed ll degree and/or increasing rotational speed decreased the percentage of particles visible on
top of the bed.
Results can be up-scaled to industrial conditions in cement rotary kilns and show that even relatively large
fuel particles will predominantly be covered by raw material after less than 30 s in the rotary kiln. This affects
the heating and combustion mechanisms for the fuel particles.
2011 Elsevier B.V. All rights reserved.
1. Introduction
Cement production is energy intensive with an energy usage of
approximately 3 MJ per kg cement clinker produced [1]. Traditionally,
cement production has mainly depended on fossil fuels such as coal,
oil and natural gas. Due to erce competition in the cement market,
increasing fossil fuel prices and environmental concerns, many
cement producers have increased the utilization of alternative fuels
as a substitute for fossil fuels in order to achieve the most economic
fuel mix. In this context, alternative fuels cover all non-fossil fuels
and waste from other industries. Popular alternative fuels in the
cement industry are tire-derived fuels, biomass residues and different
commercial and industrial wastes.
It is attractive to feed coarse, solid alternative fuels into the
material inlet end of cement rotary kilns in order to save expenses for
shredding of the fuels to smaller particle sizes and to increase fuel
exibility. High temperatures in the rotary kiln and material retention
times of several minutes provide good combustion conditions.
Rotary kilns are widely used in the cement industry where they are
well suited for the cement clinker burning processes that require high
temperatures at near-atmospheric pressure. The basic purpose of a
rotary kiln is to a) transfer heat from the kiln gasses and walls to the
solid particles in the rotary kiln bed and b) mix the solid particles in
order to facilitate the cement clinker reactions. Because the particles
on the outside of the bed insulate those on the inside, the rate of heat
transfer is directly related to the rate of mixing of particles within the
bed. Much research has been done to describe mixing of uniform
particles in rotary kilns or rotary drums [25]. Less information is
available about the mixing of particles with different sizes, shapes and
densities. This is relevant when large alternative fuel particles are
mixed with raw material particles of much smaller particle size. The
degree of mixing of these fuel particles is important since their
burnout behavior depends on mass and heat transfer. Most of the
current research about mixing of particles with different sizes has
been made with spherical balls of different materials, and typically
with size ratios of less than 5 [613].
This article studies the mixing of different fuel particles and
cement raw materials in order to predict combustion behavior. The
study was carried out under ambient conditions in a pilot scale rotary
kiln.
1.1. Bed behaviour
As a rotary kiln turns on its axis the solid bed inside is subjected to
transverse motion. The bed motion can take many forms, depending on
variables such as rotational speed, kiln diameter, ll degree, bed/wall
friction and bed particle characteristics. For conditions relevant for
clinker nodulization in cement rotary kilns, the characteristic bed
motion is the rolling bed motion [14]. This article will therefore focus on
therollingmode. It shouldbenotedhowever, that other bedmotions are
Powder Technology 210 (2011) 273280
Corresponding author at: FLSmidth A/S, Vigerslev All 77, DK-2500 Valby,
Denmark. Tel.: +45 36181000; fax: +45 36182647.
E-mail address: arni@smidth.com (A.R. Nielsen).
0032-5910/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2011.03.029
Contents lists available at ScienceDirect
Powder Technology
j our nal homepage: www. el sevi er. com/ l ocat e/ powt ec
194
possible, particularly in the material inlet end of the rotary kiln where
the partially calcined raw material particles enter with particle sizes
typically in the order of 590 m with average sizes around 20 m.
Heneinet al. have performedextensive work to establishbedbehavior
diagrams for a number of compounds [15]. This type of bed behavior
diagrams is useful to predict mode of bed motion based on bed depth and
Froude number (rotational speed). The Froude number is here denedas:
Fr =

2
R
g
=
2n
60

2
R
g
1:1
where has the unit rad/s and n has the unit min
1
. R is the drum
radius and g is the gravitational acceleration in m/s
2
. Mellmann
investigated different operating regimes based on the Froude number
[16]. He found that rolling beds typically occur for Froude numbers
between 10
4
bFrb10
2
for ll degrees higher than 10%. This result is
in good correlation with Froude numbers reported by Henein et al.
from experiments in a rotary drum with limestone and sand [15].
Henein et al. found useful correlations for scale up. The bed behavior
is not only characterized by bed ll degree and Froude number for a
given material. Particle size and shape must also be considered. For
particles with the same shape and for rotary kilns with the same
diameter and ll degree (and same bed/wall friction), the bed motion
from one particle size to another can be described as [15]:
Fr
1
D
d
p
!
1=2
1
= Fr
2
D
d
p
!
1=2
2
: 1:2
This correlation is also valid for scaling of bed behavior from one
drum diameter to another, for the same particle shape and size, and
has been used in this study to estimate the relevant rotational speeds
when scaling from an industrial rotary kiln to a pilot scale rotary kiln.
1.2. Fuel particle mixing
Cement rotary kilns are always slightly inclined with the raw
materials entering in the upper end and owing by gravity and kiln
rotation to the lower end. A large fuel particle will spend some of its
time in the active layer, on the bed surface, and the remainder of the
time in the passive layer, belowthe bed surface. The time that the fuel
particle will be present in the active layer relative to the time in the
passive layer, depends on a number of parameters, such as bed ll
degree, rotational speed, raw material characteristics, fuel particle
size, shape and density.
Although mixing occurs in both the axial and transverse direction,
mixing in the transverse direction is much more rapid [17]. This is true
for mixing in both industrial rotary kilns as well as in horizontal,
rotary drums.
The theoretical probability, P, of a large fuel particle to be present
on top of the bed rather than being buried in the bed may be described
as proportional to the volume occupied by the fuel particles relative to
the total available volume:
P
1

V
fuel
V
total

V
fuel
V
fuel
+ V
raw material

V
fuel
V
fuel
+

4
LD
2
F

100% 1:3
where V
fuel
is the volume of each fuel particle multiplied by the
number of fuel particles in the bed, L and D are the kiln length and
diameter, respectively, and F is the bed ll degree. is the sphericity
which is a measure for the non-sphericity of the fuel particle, dened
as:
=
Surface of sphere
Surface of particle

of same volume
: 1:4
With this denition, =1 for spherical particles and 0bb1 for all
other particle shapes. The sphericity is included because the shape of
the particles is likely to affect the degree of visibility, because for
particles of the same volume, the external particle surface area
increases when the sphericity decreases. This may be illustrated with
the three particle shapes shown in Fig. 1. All three particles have a
volume of 1 cubic unit, but their external surface areas are quite
different depending on their shape. The surface area increases with
decreasing sphericity and thus the particle is more likely to be fully or
partly visible in the bed.
Eq. (1.3) includes the kiln dimensions, bed ll degree, volume and
sphericity of the fuel particle. The probability increases as the fuel
volume increases. However, parameters such as rotational speed, fuel
density and raw material density are not included, even though they
are likely to have an effect. Attempts have been made to correlate
experimental values for mixing with the operating parameters and
properties of granular material. Sherrit et al. provide a review of such
correlations [17]. However, these correlations are based on mixing of
particles with the same or similar particle sizes. A similar correlation
can be derived from experiments to yield a correlation of the form:
P =
1

V
fuel
V
fuel
+

4
D
2
LF
b

Fr
c


raw material

fuel
!
d

100% 1:5
where a, b, c and d are the experimentally determined tting
parameters. Fr is the Froude number which includes the rotational
speed.
raw material
and
fuel
are the density of raw material and fuel
particles, respectively.
2. Material and methods
2.1. Experimental setup
A pilot scale rotary kiln with inner diameter and length of 50 cm
was used for the experiments, see Fig. 2. The kiln walls were made of a
deck plate with a thickness of 5 mm, and a rough inner surface. The
two end walls consisted of 600 mm diameter plexi glass plates with a
thickness of 5 mm, which was mounted with buttery screws for easy
removal during loading and unloading of the pilot scale rotary kiln.
The pilot scale rotary kiln was placed on rollers connected to a motor
with variable speed. The rotary kiln was positioned horizontally with
no inclination.
2.2. Materials
The experiments were all made with cement raw materials with
physical and chemical properties as shown in Table 1. The raw
material has a particle size that is signicantly smaller than that of the
fuel particles.
The 15 fuel particles used in the experiments are presented in
Table 2.
These fuel particles all have different shapes, sphericities, sizes,
volumes, weights and solid densities, given in the table per particle.
The sphericity is dened in Eq. (1.4). As indicated by Table 2 some of
the fuel particles have signicantly larger dimensions than the raw
material, and the size ratio may be as high as 10
4
. The size ratio is,
however, realistic, since solid fuels used in industrial cement rotary
kilns may also have dimensions up to 12 m, whereas the cement raw
material particle sizes generally are below 100 m.
2.3. Methods
The desired ll degree is obtained by lling the pilot scale rotary kiln
with the desired volume of rawmaterial. The ll degree is here dened
as the percentage of the kiln cross sectional area that is covered by raw
274 A.R. Nielsen et al. / Powder Technology 210 (2011) 273280
195
material. A xed number of fuel particles are placed on top of the raw
material, along the center line: 20 fuel particles were used in each
experiment, except for the experiments with the largest fuel particles,
sample no. 9 and no. 12 in Table 2. For these experiments with the two
largest fuel particles, only 10 fuel particles were used. The rotation is
started with the desired rotational speed. The rotational speed used in
the experiments is variedbetween3and16 rpm. It shouldhere benoted
that a rotational speed of 13 rpmin the experimental setup corresponds
to approximately 5 rpmin anindustrial rotary kiln with a 4 mdiameter,
if the scaling correlations found by Henein et al. are used [15].
A video camera was used to record the mixing in order to
document the experiments. At time intervals of 30 s the rotation is
stopped and the number of fuel particles visible on top of the bed are
counted. Then the rotation is started again. This startstop cycle is
repeated until there are only randomchanges in the number of visible
particles, or for at least 180 s. All fuel particles visible on the top of the
bed are included, even if a signicant fraction of the fuel particle is
buried in the bed.
2.4. Assumptions and uncertainties
This section discusses the main differences between the pilot scale
rotary kiln and industrial rotary kilns. The discussion presents
assumptions that have to be made in order to account for un-
certainties when results may be transferred fromthe pilot scale rotary
kiln and to industrial scale conditions.
1. The experiments are made at ambient temperature. It is assumed
that the fuel particle/raw material mixing at ambient temperature
is the same as at high temperature conditions. This assumption is
based on particle characteristics such as density which is virtually
independent on temperature. An uncertainty, however, is the
surface properties of the particles which may be quite different at
ambient and high temperature conditions. At high temperatures
the raw materials may stick to the fuel particles and e.g. affect
sphericity and owability of the fuel particles.
2. The pilot scale rotary kiln is not inclined. It is assumed that the lack
of material transport in the axial direction has no signicant effect
on the overall mixing process since mixing in the transverse plane
is dominating [17].
3. It is assumed that the dimensions of the pilot scale rotary kiln are
sufciently large to allow up-scale to industrial rotary kilns
without introducing too much uncertainty. This assumption is
based on the fact that several researchers have used similar rotary
drums to simulate industrial conditions, and in many cases
experimental results have been successfully validated against
published full scale data [15,16].
4. It is assumed that Henein et al.'s scaling principles of rotational
speed for rolling beds can be applied for the conditions used in
these experiments [15]. This assumption is conrmed by visual
observations of bed motion prior to the experiments, where the
rolling motion is seen to be dominating.
5. It is assumed that the bedwall friction coefcient in the pilot scale
rotary kiln is of comparable size as to industrial kilns: The steel wall
used in the pilot scale rotary kiln has a rough surface. In addition,
the kiln walls are covered by a thin layer of raw material that
adheres to the surface. This is assumed to be comparable to an
industrial kiln where the wall is typically covered by a coating layer
consisting of raw materials.
3. Results and Discussion
3.1. General observations
The mixing was generally observed to be a fast process that was
completed in less than 30 s, or after a few kiln rotations. After this
initial mixing, a steady state was reached where the percentage of
visible particles did not change signicantly. This observation is in
correspondence with Cantelaube et al. who also reported that the
mixing process of large particles appeared very quickly after the start
of drum rotation, typically in less than one drum revolution [10].
Fig. 1. External surface areas and sphericities for three particles with the same volume.
Fig. 2. Sketch of pilot scale rotary kiln used in the experiments.
275 A.R. Nielsen et al. / Powder Technology 210 (2011) 273280
196
The rotational velocity is varied from 3 rpm to 16 rpm, correspond-
ing to approximately 1 to 6 rpm in an industrial, 4 m diameter rotary
kiln according to the scaling correlations found by Henein et al. [15]. The
bedbehavior shouldbeinthe rollingregime under theseconditions. The
rolling bed behavior was also observed to be dominant.
Henein et al. reported that the dynamic angle of repose should be
between 30 and 50 for cement raw materials in the rolling motion
[15]. They also reported that the angle of repose is independent of ll
degree and rotational speed when the rolling motion dominates [15].
In the experiments the angle of repose was generally measured to
be between 30 and 40, and thus in good correspondence with the
ndings of Henein et al. [15].
All experiments have been repeated three times. The repeatability
is generally good with standard deviations of the experiments
generally within 48%.
Based on all the experimental data, optimal values for the tting
parameters a, b, c and d used in Eq. (1.5) were determined. The optimal
tting parameters are shown in Table 3.
The visibility predicted by the correlation will be shown graphically
in Figs. 4 to 7.
3.2. Effect of fuel particle size and shape
The effect of particle size and shape on the mixing efciency has
been investigated for ll degrees between 5% and 30% and rotational
speeds between 3 rpm and 16 rpm. 12 samples with different sizes
and shapes were used in the experiments, see Table 2 The sphericity is
dened in Eq. (1.4). As indicated in Table 2 some of the fuel particles
have signicantly larger dimensions than the raw material, and the
size ratio may be as high as 10
4
. The size ratio is, however, realistic,
since solid fuels used in industrial cement rotary kilns may also have
dimensions up to 12 m, whereas the cement raw material particle
sizes generally are below 100 m.
See Table 2 for details. An example for a ll degree of 10% and a
rotational speedof 13 rpmis showninFig. 3. It is observedthat particles,
regardless of size and shape, decrease rapidly from 100% percent
visibility to signicantly lower values. The mixing takes place within the
rst 30 s, corresponding to 18 kilnrotations. After this, the particles are
to a large degree covered by rawmaterial and only appear on top of the
bed randomly.
It is noted that the large, rectangular samples numbers 9 and 12 are
signicantly more visible on top of the bed: After 30 s approximately
47% and 58% of these particles are still visible, respectively. And these
percentages are more or less constant with time. On the contrary, the
smallest particles which are the spherical ceramic particles denoted
sample 4, drop from 100% to 0% during the rst 30 s. Thereafter, the
small particles are not observedmore ontopof the bed, during the 180 s
that the experiment runs. The largest and most visible fuel particles
numbers 9 and 12 are also the least spherical particles, whereas the
smallest and least visible fuel particles are also the only truly spherical
particles used in the experiments. These results indicate that not only
the fuel particle dimensions but alsothe fuel particle shape, expressedas
sphericity, is important for the visibility above the bed. This effect of
sphericity is further conrmed by comparing samples numbers 1 and
8 which have similar volumes of 57 cm
3
and 66 cm
3
, respectively, but
quite different sphericities of 0.95 and 0.67. It is observed that the least
Table 1
Composition and physical properties of cement raw material used in the
experiments.
Chemical composition
CaCO
3
78.4 wt.%
SiO
2
13.5 wt.%
Fe
2
O
3
2.3 wt.%
Al
2
O
3
3.3 wt.%
Minor components 2.6 wt.%
Bulk density 1200 kg/m
3
Number mean particle size 18 m
Sieve interval, m
05 21.4 wt.%
510 13.3 wt.%
1020 19.0 wt.%
2030 12.5 wt.%
3060 18.9 wt.%
6090 8.2 wt.%
90400 6.6 wt.%
Table 2
Fuel particles used in the experiments. Size, volume and weight are per particle. The
density is the solid density.
Sample
No.
Particle
type
Shape/
sphericity
Size [cm] Volume
[cm
3
]
Weight
[g]
Density
[kg/m3]
1 Charcoal Oval/0.95 5.55.53.0 56.8 45 792
2 Charcoal Oval/0.95 5.52.83.0 29.0 23 792
3 Charcoal Irregular/0.89 2.82.83.0 10.1 8 792
4 Ceramic Spherical/1.0 D=0.6 0.11 0.32 2829
5 Wood Cubic/0.81 222 8.0 3.3 470
6 Wood Cylindric/0.87 D=2.2, L=2 7.6 5.1 665
7 Wood Cylindric/0.84 D=2.2, L=4 15.2 10 665
8 Wood Rectangular/0.67 1.84.58.1 66.0 31 470
9 Wood Rectangular/0.56 1.84.516.5 134.0 78 470
10 Tire
rubber
Rectangular/0.78 2.52.04.0 20.0 22 1100
11 Tire
rubber
Rectangular/0.75 2.55.06.0 75.0 82.5 1100
12 Tire
rubber
Rectangular/0.59 2.58.017.0 340.0 374 1100
13 Plastic Cylindric/0.73 D=1.3, L=5.3 7.0 2.4 341
14 Plastic Cylindric/0.73 D=1.3, L=5.3 7.0 4.0 569
15 Plastic Cylindric/0.73 D=1.3, L=5.3 7.0 7.7 1095
Table 3
Fitting parameters used in the correlation between percent visible particles and
experimental data.
a b c d
1.50.2 1.20.3 0.090.01 0.770.2
0 30 60 90 120 150 180
Time [sec.]
0
20
40
60
80
100
P
a
r
t
i
c
l
e
s

v
i
s
i
b
l
e

a
b
o
v
e

b
e
d

[
%
]
Sample No 1: Charcoal
Sample No. 2: Charcoal
Sample No. 3: Charcoal
Sample No. 4: Plastic
Sample No. 5: Wood
Sample No. 6: Wood
Sample No. 7: Wood
Sample No. 8: Wood
Sample No. 9: Wood
Sample No. 10: Tyre rubber
Sample No. 11: Tyre rubber
Sample No. 12: Tyre rubber
Fig. 3. Example of raw data: Percent particles visible above bed as a function of time.
10% ll degree and 13 rpm.
276 A.R. Nielsen et al. / Powder Technology 210 (2011) 273280
197
spherical particle is the most visible. However, the density of the two
samples is not identical and this may affect the visibility as well. Inorder
to determine the effect of shape with greater accuracy, it is required to
conduct more experiments with fuel particles of similar volumes and
densities but different shapes.
The experiments have been repeated for different ll degrees and
rotational speeds and the tendency is the same under all conditions:
Fuel particles with large dimensions have a higher probability to be
visible on top of the bed relatively to smaller particles that are to a
higher degree buried in the bed. This geometric effect on the degree of
visibility was also predicted by Eqs. (1.3) and (1.5).
Fig. 4 shows the visibility of all the different fuel particles at steady
state, sample no. 1 to sample no. 12, as a function of the fuel particle
volume. Steady state is here dened as the time interval between 30 s
and 180 s, where the percent visibility is relatively constant.
The small ceramic spheres, sample no. 4, had the smallest volume
of 0.11 cm
3
per particle, and has a steady state visibility of close to 0%.
Several of the tested fuel particles had volumes around 10 to 60 cm
3
,
but different sphericities and densities. The steady state visibility is
still low, around 37%. For fuel particles with volumes larger than
100 cm
3
the percent visibility increases signicantly, up to 50% for the
largest particle with a volume of 340 cm
3
. The standard deviation
varied from 0% to 8% during the experiments, and was observed to be
greatest for the largest fuel particles.
Fig. 4 also shows the estimated probability for the particles to be
visible based onEqs. (1.3) and(1.5). EventhoughEq. (1.3) does not take
the effect of density and rotational speed into account, a fairly good
consistency is seenbetweenthe theoretical andexperimental values, for
volumes up to around 100 cm
3
. But for the largest particles, Eq. (1.3)
under-estimates the probability: This is particularly observed for the
particle with a volume of 134 cm
3
, where the observed visibility is 43%
but the theoretical probability is only 21%. The reason why Eq. (1.3) fails
to give a good estimation of the theoretical visibility for the largest
particles may be that it becomes increasingly difcult for the largest
particles to be fully buried in the bed, as their dimensions increase. Thus
the degree of visibility will deviate more and more from the theoretical
predictions as the particle dimensions increases.
Eq. (1.5) whichincludes correlations for rotational speed and density
and the experimentally determined tting parameters presented in
Table 3 give a good consistency between theoretical and experimental
values. The greatest deviation is for the particle with a volume of 57 cm
3
which deviates 4% from the experimental value. But it is also observed
that sample no. 10 and sample no. 1 with volumes of 20 cm
3
and 57 cm
3
,
respectively, lie ona slightly lower visibility thanthe other particles with
volumes intherange 1070 cm
3
. Areasoncanbethat these twoparticles
have a high density, see Section 3.5. However, as indicated by the
standard deviations, the reason may also simply be due to uncertainties.
The observations regarding fuel particle size and the degree of
visibility above the bed is ingoodcorrespondence withthe theory about
void lling presented by Savage and Lun [18]: During mixing of several
layers of particles in a bed, void spaces between the particles are
undergoing continual random changes. If a void space is large enough,
then a particle from the above layer can fall into it as the layers move
relative to eachother. The probability of nding a voidspace that a small
(raw material) particle can fall into is larger than the probability for
nding a void space that a large (fuel) particle can fall into. Thus, this
gravity-induced size-dependent void lling mechanism creates a
tendency for small particles to end at the bottom and large particles to
end at the top.
The observations may also be considered as a pure geometric effect
where increasing dimensions of the fuel particles imply a decreasing
probability that the fuel particles will be fully covered by the raw
material bed.
3.3. Effect of bed ll degree
The rawmaterial bedll degree has beenvariedfor all samples inthe
following range: 5%, 10%, 15% and 20%. This range of ll degrees covers
the situation in industrial cement rotary kilns, where ll degrees
typically are 815%.
Fig. 5 shows the effect of ll degree for tire rubber particles of three
different sizes and thus with three different volumes, as specied in
Table 2. For all three particle sizes are the percent visibility observed
to decrease with increasing ll degrees.
The large particles are around 80% visible at the 5% ll degree. At
10% ll degree and 15% ll degree the percentages are around 50% and
40%, respectively.
It was found that the ll degree is an important parameter for the
percentage of fuel particles that is present on top of the bed. Especially
small particles will, regardless of shape, rapidly be buried in the bed.
Larger particles will, due to their proportions, only be partly buried in
the bed whereas part of the particles will be sticking out in the
freeboard above the bed.
Thedegreeof visibilitypredictedbyEq. (1.5) is shownwithsolidlines
for each of the three particle sizes. The correlation ts the experimental
data well with deviations generally less than 3%.
3.4. Effect of bed rotational speed
Fig. 6 shows the effect of rotational speed on the mixing of tire
rubber particles, samples no. 10, 11 and 12 in Table 2. The rotational
speed is expressed in terms of the dimensionless Froude number. It is
observed that the percentage of visible particles decreases when the
Froude number increases: Approximately 28% of the medium sized
particles are visible at the lowest Froude number of 0.0024, whereas
only 9% are visible at the highest Froude number of 0.0702. For the
largest particles, 68% of the particles are visible at the lowest Froude
number and 48% are visible at the highest Froude number. The smallest
particles, however, do not show any particular dependence of Froude
number. The same tendencies were also observed for other ll degrees
and other particles.
The degree of visibility predicted by Eq. (1.5) is shown with solid
lines for each of the three particle sizes. The correlation ts the
experimental data well with deviations generally less than 3%.
0.1 1 100 10 1000
Particle volume [cm
3
]
0
20
40
60
P
a
r
t
i
c
l
e
s

v
i
s
i
b
l
e

a
b
o
v
e

b
e
d

[
%
]
Experimental data
Visibility predicted by equation (1.3)
Visibility predicted by equation (1.5)
Fig. 4. Percent visible particles at steady state versus particle volume. 10% ll degree,
13 rpm. Eq. (1.3) is predictive while Eq. (1.5) has been adapted on the experimental
data.
277 A.R. Nielsen et al. / Powder Technology 210 (2011) 273280
198
The effect of the rotational speed on percentage of visible particles
can be explained by the fact that a high rotational speed is equal to
more bed revolutions and thus a more efcient mixing process
relative to a lower rotational speed [15]. It should, however, be noted
that the percent visibility of particles seems to stabilize at a certain
percentage for each rotational speed. These differences in steady state
visibility at different rotational speeds may be due to the changes in
centrifugal forces acting on the fuel particles which may affect the fuel
particle distribution in the raw material bed.
Thomas and Cantelaube et al. also studied the effect of rotational
speed on the segregation of particles up to 3 mm and 20 mm in
diameter, respectively [11,13]. They found no effect of the rotational
speed for these particle sizes, which are in correspondence with the
observations in this study, for the smallest particles.
3.5. Effect of fuel density
The effect of particle density was studied by lling plastic containers
of identical size and shape with materials of different weight. Thus, the
outer dimensions of the particles were identical, but the densities were
varied from low, medium and high, see Table 2 for details.
Fig. 7 shows the percentage of visible particles versus the raw
material/fuel particle density ratio for ll degrees of 5%, 10% and 15%,
and rotational speeds of 3, 10 and 16 rpm. Standard deviations on the
results presented in Fig. 7 were between 2% and 6%. It is observed that
the greatest effect of particle density appears with the 5% ll degree
and 3 rpm: The low-density particles stabilize at a level around 45%
visible particles, the medium-density particles stabilize at a level
around 43% visible particles, whereas the high-density particles are
only 28% visible. The same effect of particle density is also observed at
10% ll degree and 3 rpm. It is generally observed that the lowest
rotational speed of 3 rpm results in the highest percentage of visible
particles. The effect of density is observed to decrease as the ll degree
and/or rotational speed increases.
Fig. 7 also shows the values predicted fromEq. (1.5), with solid lines
and symbols. Eq. (1.5) predicts the visibility reasonably well, except for
the two cases with the lowest rotational speed of 3 rpm: In these two
cases the deviations are as high as 20%, which is a signicantly higher
deviation than observed for any other combinations of density,
rotational speed, ll degree and fuel particle size and shape. There is
thus a poor relationship between the tting parameter for density, and
some combinations of operational parameters.
The effect of particle density on the percent visibility is in good
correspondence with the push-away theory presented by Tanaka,
0 5 10 15 20
Fill degree [%]
0
20
40
60
80
P
a
r
t
i
c
l
e
s

v
i
s
i
b
l
e

a
b
o
v
e

b
e
d

[
%
]
Particle surface area [cm
2
]
46
115
397
Fig. 5. Percent visible tire rubber particles versus bed ll degree for three different
particle sizes. Predicted values are shown with solid lines and experimental values are
shown with symbols. 10 rpm.
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Fr
0
20
40
60
80
P
a
r
t
i
c
l
e
s

v
i
s
i
b
l
e

a
b
o
v
e

b
e
d

[
%
]
Particle surface area [cm
2
]
46
115
397
Fig. 6. Percent visible tire rubber particles versus Froude number for three different
particle sizes. 10% ll degree. Predicted values are shown with solid lines and
experimental values are shown with symbols.
1 1.5 2 2.5 3 3.5 4
Particle ratio [

Raw material
/

Fuel
]
0
10
20
30
40
50
P
a
r
t
i
c
l
e
s

v
i
s
i
b
l
e

a
b
o
v
e

b
e
d

[
%
]
5% fill degree, 3 rpm
5% fill degree, 10 rpm
5% fill degree, 16 rpm
10% fill degree, 3 rpm
10% fill degree, 10 rpm
10% fill degree, 16 rpm
15% fill degree, 10 rpm
Fig. 7. Percent particles visible above bed at steady state versus particle density.
Estimated values are shown with solid lines and symbols while experimental values are
shown with symbols only.
278 A.R. Nielsen et al. / Powder Technology 210 (2011) 273280
199
which states that dense particles are more likely to push away other
particles and sink to the bottom [19].
4. Practical implications
Eq. (1.5) is applied to conditions of an industrial rotary kiln in
order to estimate the effect of fuel particle volume and sphericity on
the percent visibility above the raw material bed. Typical values for a
cement rotary kiln are shown in Table 4.
Two selected fuels, tire rubber and wood are studied. The fuel
densities will decrease rapidly during the heating and devolatiliza-
tion process in the material inlet where raw material temperatures
and gas temperatures are typically 900 C and 1100 C, respectively.
It is assumed that the fuel char densities are representative during
the mixing process. The char density of tire rubber is assumed to
be 340 kg/m
3
and the char density of wood is assumed to be 160 kg/
m
3
.
Fig. 8 shows the percent visibility of tire rubber char and wood char
particles as a function of the fuel particle volume and for three different
fuel particle sphericities of 0.5, 0.75 and 1.0, respectively. The volume of
fuel particles is assumed to be distributed over the rst 10 m of the
material inlet end of the rotary kiln, where the combustion is assumed
to take place. The percent visibility is observed to increase linearly with
the fuel particle volume. It is interesting to note that the sphericity of
the fuel particles has a great effect on the percent visibility: A pure
spherical fuel particle (=1.0) is least visible due to its compact shape
which leads to a high degree of fuel coverage by raw materials. As the
fuel particle sphericity decreases, the percent visibility increases due to
the larger dimensions of the fuel particles. The increasing dimensions
of fuel particles of the same volume but decreasing sphericity are
illustrated in Fig. 1. The tendency with sphericity and visibility is the
same for tire rubber char and wood char but it is observed that the
wood char is considerably more visible than tire rubber char due to its
lower density.
An important practical implication for combustion processes in
industrial rotary kilns can be concluded to be the shape of the fuel
particles, because low sphericity fuel particles are expected to be in
better contact with the gas phase, which may improve the combustion
process. Also the fuel particle density will have an effect on the gas
fuel contact.
Full scale experience at two different cement plants with com-
bustion of whole automobile tires and tire chips have shown that the
tendency for undesirable effects on process stability and cement
clinker quality due to reducing conditions in the rawmaterial bed was
less severe with whole tires relative to the situation with tire chips
[20,21]. These results indicate that the smaller tire chips are to a
higher degree buried in the rawmaterial bed and create local reducing
conditions. The larger whole tires will to a larger degree stay above
the bed and be exposed to higher temperatures as well as to improved
oxygen mass transfer from the freeboard gas. The industrial
experience is in line with the experimental results in this study
regarding the tendency for large fuel particles to be in better contact
with the freeboard gas above the raw material bed relative to smaller
fuel particles.
5. Conclusions
During experiments in a pilot scale rotary kiln distribution of large
solid fuel particles in cement raw materials were studied. The results
were evaluated as the percentage of fuel particles visible above the
bed at specic time intervals.
The fuel particles were to a large degree buried in the cement raw
materials within less than 30 s. After this, the percentage of visible fuel
particles stabilized at a specic level, depending on fuel particle size.
There was a clear tendency that large fuel particles were more visible
than smaller fuel particles. Fuel particle dimensions and sphericity are
important parameters for the degree of visibility.
Low-density fuel particles were more present above the bed than
high-density fuel particles which were to a higher degree buried in
the bed. The inuence of density was particularly pronounced at low
ll degrees and/or low rotational speeds.
The bed ll degree is of major importance for the percentage of
visible fuel particles in the bed. Small bed ll degrees favor the
number of fuel particles visible above the bed.
There is a tendency that an increase in rotational speed reduces the
number of fuel particles visible above the bed. However, the effect of
rotational speed was not as signicant as the effect of bed ll degree.
A correlation for the percentage of visible fuel particles and the
operating parameters was presented and compared with the experimen-
tal results. The correlation did generally predict the degree of visibility
within 3%. However, the correlation failed to give good predictions in the
cases with the lowest rotational speed of 3 rpm and density as a variable
parameter.
Acknowledgments
The work described in this article is part of a research platform on
future cement technology nanced by The Danish National Advanced
Technology Foundation, Technical University of Denmark and FLSmidth
A/S.
Table 4
Typical data for a modern cement rotary kiln (FLSmidth one string ROTAX 2 kiln).
Rotary kiln
Inner diameter, m 4.35
Total rotary kiln length, m 51
Material inlet section length, m 10
Rotational speed, min
1
4
Angle of tilt, 3
Raw material ow, tpd (kg clinker/s) 3500 (40.51)
Mean raw material residence time, min 11
Mean raw material ll degree,% 11
Raw material density, kg/m
3
1200
0 0.4 0.8 1.2 1.6
Particle volume [m
3
]
0
20
40
60
80
100
P
a
r
t
i
c
l
e
s

v
i
s
i
b
l
e

a
b
o
v
e

b
e
d

[
%
]
Tyre rubber, Sphericity = 0.5
Tyre rubber, Sphericity = 0.75
Tyre rubber, Sphericity = 1.0
Wood, Sphericity = 0.5
Wood, Sphericity = 0.75
Wood, Sphericity = 1.0
Fig. 8. Estimated degree of visibility of tire rubber char and wood char particles as a
function of fuel particle volume and sphericity. Estimated by Eq. (1.5).
279 A.R. Nielsen et al. / Powder Technology 210 (2011) 273280
200
References
[1] M. Schneider, VDZ Activity Report 20072009, Germany, 2009.
[2] G.R. Woodle, J.M. Munro, Particle motion and mixing in a rotary kiln, Powder
Technol. 76 (1993) 241245.
[3] A. Santomaso, M. Olivi, P. Canu, Mixing kinetics of granular materials in drums
operated in rolling and cataracting regime, Powder Technol. 152 (2005) 4151.
[4] D.R. van Puyvelde, Comparison of discrete elemental modelling to experimental
data regarding mixing of solids in the transverse direction of a rotating kiln, Chem.
Eng. Sci. 61 (2006) 44624465.
[5] M. Kwapinska, G. Saage, E. Tsotsas, Mixing of particles in rotary drums a
comparison of discrete element simulations with experimental results and
penetration models for thermal processes, Powder Technol. 161 (2006) 6978.
[6] S. Tallon, C.E. Davies, In-situ monitoring of axial particle mixing in a rotating drum
using bulk density measurements, Powder Technol. 186 (2008) 2230.
[7] A.V. Orpe, D.V. Khakhar, Scaling relations for granular ow in quasi-two-
dimensional rotating cylinders, Phys. Rev. E 64 (031302) (2001) 113.
[8] S.K. Hajra, D.V. Khakhar, Sensitivity of granular segregation of mixtures in quasi-
two-dimensional uidized layers, Phys. Rev. E 031304 (2004) 14.
[9] E. Clment, J. Rajchenbach, J. Duran, Mixing of a granular material in a
bidimensional rotating drum, Europhys. Lett. 30 (1995) 712.
[10] F. Cantelaube, D. Bideau, Radial segregation in a 2d drum: an experimental
analysis, Europhys. Lett. 30 (1995) 133138.
[11] F. Cantelaube, D. Bideau, S. Roux, Kinetics of segregation of granular media in a
two-dimensional rotating drum, Powder Technol. 93 (1997) 111.
[12] M. Alonso, M. Satoh, K. Miyanami, Optimum combination of size ratio, density
ratio and concentration to minimize free surface segregation, Powder Technol. 68
(1991) 145152.
[13] N. Thomas, Reverse and intermediate segregation of large beads in dry granular
media, Phys. Rev. E 62 (2000) 961974.
[14] A.A. Boateng, Rotary Kilns, Elsevier Inc., USA978-0-7506-7877-3, 2008.
[15] H. Henein, J.K. Brimacombe, A.P. Watkinson, Experimental study of transverse bed
motion in rotary kilns, Metall. Trans. B 14B (1983) 191205.
[16] J. Mellmann, The transverse motion of solids in rotating cylinders forms of
motion and transition behaviour, Powder Technol. 118 (2001) 251270.
[17] R.G. Sherritt, J. Chaouki, A.K. Mehrotra, L.A. Behie, Axial dispersion in the three-
dimensional mixing of particles, Chem. Eng. Sci. 58 (2003) 401415.
[18] S.B. Savage, C.K.K. Lun, Particle size segregation in inclined chute ow of dry
cohesionless granular solids, J. Fluid. Mech. 189 (1988) 311335.
[19] T. Tanaka, Segregation models of solid mixtures composed of different densities
and particle sizes, Ind. Eng. Chem. Process Des. Develop. 10 (1971) 332340.
[20] H.-E. Jannerup, Personal Communication, FLSmidth A/S, Valby, Denmark,
September 2009.
[21] J.E. Lund, Personal Communication, Cemex Philippines, Tinaan, Naga, Cebu,
Philippines, June 2010.
280 A.R. Nielsen et al. / Powder Technology 210 (2011) 273280
201
,MKL8IQTIVEXYVI 6IPIEWI SJ 73 JVSQ 'EPGMRIH 'IQIRX
6E[ 1EXIVMEPW
Anders R. Nielsen,' Morten . Iarsen, Ieter Glarborg, and kim LamJohansen
IISmidth A/S, Vigerslev Alle , DK-2;00 Valby, Denmark
Department of Chemical Ingineering, CHICResearch Centre, Technical University of Denmark (DTU), DK-2800 Iyngby, Denmark
ABS1RAC1: During combustion of alternative fuels in the material inlet end of cement rotary kilns, local reducing conditions may
occur and cause reductive decomposition of sulfates from calcined cement raw materials. Decomposition of sulfates is problematic
because it increases the gas-phase SO concentration, which may cause deposit formation in the kiln system. In this study, the release
of sulfur from calcined cement raw materials under both oxidizing and reducing conditions is investigated. The investigations
include thermodynamic equilibrium calculations in the temperature interval of 800 1;00 C and experiments in a tube furnace
reactor in the temperature interval of 900 1100 C. The investigated conditions resemble actual conditions in the material inlet end
of cement rotary kilns. It was found that the sulfates CaSO , K SO , and Na SO were all stable under oxidizing conditions but
began to decompose under reducing conditions. Particularly, CaSO was sensitive to reducing conditions. The sulfur release was
most signi cant if the gas atmosphere frequently shifted between oxidizing and reducing conditions. An increasing temperature from
900 to 1100 Cunder alternating oxidizing and reducing conditions was also observed to increase the sulfur release fromthe calcined
raw materials by a factor of 3, from 14 to 48.
v-2863(9'8-32
Cement production is energy-intensive, with an energy usage
of approximately 3 MJ/kg of cement clinker produced. With an
annual global cement production of 2.83 billion tons of cement,
the cement industry is responsible for approximately 2 of the
worldvs primary energy consumption.
Iuel consumption accounts for about 30 40 of the total
cement clinker production costs. Traditionally, cement produc-
tion has mainly depended upon the fossil fuels coal, oil, and
natural gas. Because of erce competition in the cement market,
increasing fossil fuel prices, and environmental concerns, cement
producers have increased the use of alternative fuels as a
substitute for fossil fuels to achieve the most economic fuel
mix. In this context, walternative fuelsx cover all nonfossil fuels
and waste from other industries. Popular alternative fuels in the
cement industry are tire-derived fuels, biomass residues, and
di erent commercial and industrial wastes.
It is attractive to use coarse, solid alternative fuels in the
material inlet end of cement rotary kilns to save expenses for
shredding of the fuels to smaller particles and to increase fuel
exibility of the system. High temperatures in the rotary kiln and
material retention times of several minutes provide good condi-
tions for fuel burnout.
Iigure 1 shows a simpli ed owdiagramand main sulfur reactions
of a rotary kiln and calciner. Preheated cement raw materials are
admitted to the calciner from the preheater (the preheater above the
calciner is not shown). Because the rawmaterials have beenpreheated
to around 800 C, only inorganic-bound sulfur is present in the raw
materials, mainly in the formof sulfates. Besides sulfur input fromthe
cement raw materals, sulfur may also be introduced with the fuels in
either calciner or rotary kiln. Sulfur from the fuel may be either
organic- or inorganic-bound, and in Iigure 1, sulfur input via the fuels
is indicated by a S in parentheses.
The calciner acts as a uidized-bed reactor where good contact
between hot gas from the rotary kiln or tertiary air duct and raw
materials from the preheater provide optimal conditions for
gas solid heat transfer and fuel combustion that drives the
endothermic calcination of limestone.
CaCO
3
s 8 CaOs CO
2
g 1
Besides calcination of limestone, the conditions in the calciner
with temperatures in the range of 800 900 C, good gas solid
mixing, and excessive amounts of CaO also favor the SO capture
reaction.
CaOs SO
2
g
1
2
O
2
g 8 CaSO
4
s 2
Because of the e cient SO capture by CaO in the calciner, SO
emissions because of sulfur oxidation in the rotary kiln are
negligible at modern cement plants equipped with a calciner.
SO emissions from cement plants are mainly due to oxidation of
pyrite, Ie S, from the cement raw materials in the early stages of
the preheater. An exception, however, are cement plants oper-
ating with a bypass, where a fraction of the kiln gases are released
to the atmosphere without being transported through the
calciner. This may result in increased SO emissions.
The partially calcined raw materials from the calciner are
admitted to the material inlet end of the rotary kiln, where the
calcination reaction is completed. In this paper, fully or partly
calcined cement raw materials will be denoted calcined raw meal.
Sulfur present in the calcined raw meal is mainly bound as
CaSO or as alkali sulfates. The calcined raw meal is then
Received: April 20, 2011
Revised: May 23, 2011
202
transported by the rotational movement and slight inclination of
the rotary kiln toward the hot burning zone, where ame tempera-
tures are in the range of 2000 C and material temperatures are
around 14;0 C. The clinker reactions take place during the gradual
heating of the calcined raw meal as it is transported through the
rotary kiln. During the gradual heating of the calcined raw meal, a
sulfate melt is formed, which facilitates the formation of alkali
sulfates, Na SO and K SO , which are more thermodynamically
stable than CaSO . CaSO will decompose at the high tempera-
tures present in the burning zone and form SO .
CaSO
4
s 8 SO
2
g
1
2
O
2
g CaOs 3
The fraction of CaSO that decomposes in the burning zone
depends upon parameters such as retention time, clinker and gas
temperatures, gas composition, and clinker nodule size. High SO
concentrations in the gas phase suppress dissociation of CaSO to
CaO and SO in the calcined raw meal.
SO formed in the rotary kiln is transported with the kiln gases
to the calciner, where the previously described sulfur capture on
CaO takes place. Thus, an internal sulfur cycle in the rotary kiln
and calciner is established. The fraction of sulfur released as SO
depends upon the sulfur/alkali ratio in the rotary kiln. High levels
of alkali metals will favor the formation of alkali sulfates rather
than the formation of free SO .
After the rotary kiln, the cement clinker is quench-cooled by
ambient air as cooling air. Sulfur in the cement clinker is mainly
bound as Na SO , K SO , or 3K SO

Na SO . However, it may
also be present as CaSO .
The main challenge with respect to alternative fuel use in the
material inlet end of rotary kilns is that the solid fuel particles will be in
physical contact with the calcined raw meal. During the fuel
devolatilization, reducing agents, such as CO, H , and/or hydro-
carbons, are formed. These reducing agents may react with minor
elements in the calcined raw meal before they are oxidized to their
ultimate combustion products, H O and CO . In addition, if the fuel
particles are fully or partly covered by calcined rawmeal, mass transfer
of oxygen to the fuel char will be hindered. Sub-stoichiometric
amounts of oxygen will lead to incomplete oxidation of the fuel char,
forming reducing agents in form of CO, H , and/or hydrocarbons.
Iigure 1. Simpli ed ow diagram of a rotary kiln, calciner, and material ows from a preheater, showing sulfur balances. Arrows indicate the ow
direction, and the principal reactions are shown.
203
It is widely accepted that Portland cement should be burnt
under oxidizing combustion conditions. The reason is that the
existence of local reducing conditions in the calcined raw meal
charge may a ect the product quality and process stability of the
kiln system. The product quality can be in uenced by calcined
raw meal components, such as Ie , being reduced to Ie . Ie
catalyzes alite (3CaO

SiO ) decomposition, the main strength-


giving component in cement. The process stability is a ected
by the increased release of sulfur from the calcined raw meal,
mainly by reductive decomposition of CaSO .
CaSO
4
s COg 8 CaOs CO
2
g SO
2
g 4
This reaction is shown in the material inlet end of the rotary kiln
(see Iigure 1). Increasing amounts of SO in the cement kiln
system are problematic because SO promotes the formation of
sulfospurrite [2(2CaO

SiO )

CaSO | and calcium sulfoalumi-


nate (3CaO

3Al O

CaSO ), some of the principal constitu-
ents of deposit buildups found in rotary kilns and kiln riser
ducts (see Iigure 1). These sulfur-containing deposit buildups
can lead to blockages that need to be removed, sometimes by a
temporary plant shutdown.
With respect to local reducing conditions caused by alternative
fuel combustion in the material inlet end of cement kilns, the
most important challenge is assumed to be process stability
because of sulfur release from the calcined raw materials and to a
lesser degree product quality. Product quality is mainly a ected
by local reducing conditions in the kiln burning zone, which is a
di erent situation.
The aim of this work is to obtain quantitative data on the
release of SO from calcined cement raw materials under condi-
tions that resemble those in the material inlet end of rotary kilns.
The purpose is to document the impact of variations in the local
stoichiometry on the sulfur release. This type of information is
useful in the prediction of e ects on sulfur release during use of
solid alternative fuels in the material inlet end of cement rotary
kilns. The work includes both thermodynamic calculations on
phase equilibria and experiments in a high-temperature tube
furnace reactor. As described in the next section, most related
knowledge is obtained under uidized-bed conditions, where
both temperatures and Ca/S ratios are typically lower than in the
material inlet end of cement rotary kilns.
v,-+,8)14)6%896) 6)%'8-327 &)8;))2 73
%2( 0-1)7832)
Absorption of SO on limestone under uidized-bed conditions
has been studiedextensively during the last fewdecades. Iven
though these investigations are not directly applicable to describe
the conditions in the material inlet end of cement rotary kilns, they
do include relevant descriptions of mechanisms for sulfur release
and capture by limestone at temperatures up to 1200 C and under
oxidizing as well as reducing atmospheres. An important point to
note however is that these investigations only include reactions
between limestone and sulfur species, whereas a full investigation of
the reactions taken place in cement rotary kilns should also include
other major components present in the calcined cement raw
materials because they may a ect the sulfur chemistry. In addition,
the Ca/S molar ratio studied under uidized-bed conditions was
typically inthe range of 11 41 which di ers fromthe conditions in
cement rotary kilns, where there is a massive excess of Ca relative to
sulfur. The Ca/S molar ratio in the calcined rawmeal that enters the
rotary kiln is typically higher than 101.
The following reaction mechanism for reductive decomposi-
tion of CaSO has been proposed
CaSO
4
s COg 8 CaSO
3
s CO
2
g ;
CaSO
3
s 8 CaOs SO
2
g
CO is believed to react directly on CaSO , giving CaSO and
CO . CO is rapidly desorbed, while CaSO disproportionates
and forms CaO and SO . CaSO is believed to be an important
reaction intermediate. The formed CaO is subsequently sul -
dized to CaS according to
CaOs SO
2
g 3COg 8 CaSs 3CO
2
g
Returning to oxidizing conditions, CaS is oxidized to CaO by the
reaction
CaSs 1;O
2
g 8 CaOs SO
2
g 8
The CaS oxidation is a rapid and very exothermal reaction, which
may lead to a temperature increase. The nal step is the formation
of CaSO by the reaction of CaO, SO , and O , which is also an
exothermal reaction
CaOs SO
2
g
1
2
O
2
g 8 CaSO
4
s 9
Hansen et al. studied phase equilibria for the SO CaO
CaSO CaS CO CO system. They performed experiments in
an electrically heated laboratory-scale xed-bed reactor developed to
simulate the changing oxidizing and reducing conditions similar to
the conditions that particles will experience in a uidized-bed
reactor. They found that any transformation between CaSO and
CaS takes place via CaO. This transformation cycle is shown in
Iigure 2, which also illustrates the competition between sulfur
capture and sulfur release. This competition depends upon para-
meters such as partial pressures of SO , CO, and CO (see the phase
Iigure 2. Sulfur transformation during periodically changing oxidizing
and reducing conditions.
1ab|e 1. Ca|cined Raw Mea| Composition Used in the
Lqui|ibrium Ca|cu|ations
calcined raw meal wt mol/kg of clinker
SiO 18.00 3.000
Al O 4. 0.4
Ie O 2.30 0.144
CaO ;.80 10.128
K O 2.;0 0.2
Na O 0.20 0.032
S total 2. 0.829
Cl 1.1 0.330
204
diagramin Iigure 3). The temperature is a very important parameter.
An increase in the temperature in the interval of 800 1200 C will
shift the equilibrium toward CaO and release more SO .
During devolatilization and combustion of alternative fuels in
an oxygen-lean environment, several reducing agents may be
formed CO, H , CH , as well as other hydrocarbons. It is well-
documented that reducing agents, such as CO and H , can
promote reductive decomposition of CaSO . Hansen
et al. studied the rate of reductive decomposition with three
reducing agents, CO, H , and CH . It was observed that the rate
of decomposition was fastest with H as a reducing agent. This is
in agreement with earlier ndings, which showed that the rate
was increased by a factor 2 or 3 when the reducing agent was H
instead of CO. Also, Diaz-Bossio et al. reported that the rate of
reductive decomposition was fastest with H relative to CO.
Hansen et al. did not detect any reductive decomposition of
CaSO with CH as a reducing agent, even though the reaction is
thermodynamically feasible. It was assumed to be because
reductive decomposition of CaSO with CH is a very slow
process that did not have su cient time to proceed during the 30
s intervals used in the experiments.
Turkdogan and Vinters also showed that elemental carbon could
be used as a reducing agent. The conversion rate of CaSO to SO
and/or CaS was reported to be determined by the rate of oxidation
of carbon to CO, which then rapidly reduced CaSO .
v8,)613(=2%1-' )59-0-&6-91 '%0'90%8-327
To provide the basis for a better understanding of the
inorganic chemistry in the rotary kiln material inlet, thermo-
dynamic equilibrium calculations were performed using the
commercially available software IactSage .0. This program uses
the principle of minimization of the total Gibbs free energy to
calculate the equilibrium composition of a chemical system with
known total elemental composition, temperature, and pressure.
-RTYX (EXE The equilibrium calculations are based on the
composition of calcined cement rawmeal and the flue gas leaving
the rotary kiln. These compositions are shown in Tables 1 and 2.
It should be noted that the calcined raw meal used in the
calculations contains a relatively high amount of chlorine and
sulfur, relative to the guidelines for a stable kiln operation.
The following input data were used (1) Chemical composi-
tions of calcined raw meal and ue gas are as speci ed in Tables 1
and 2. All elements are inserted as mol/kg of clinker. (2) It is
assumed that there is 1 kg of calcined raw meal and 0.; N m ue
gas for each kilogram of clinker produced. (3) The following
elements were included C, H, N, O, S, Cl, K, Na, Ca, Si, Ie, and
Al. (4) The ue gas is either oxidizing ( = 1.3) or reducing ( =
0.9). (;) Temperature = 800 1;00 C, with a step size of 2; C.
() Pressure = 1 atm.
Ior the equilibrium calculations, ideal gas-phase compounds,
solid-phase compounds, and liquid-phase compounds were in-
cluded from the IactSage .0 compound database.
0MQMXEXMSRW The equilibrium model has the following limita-
tions, which should be considered when comparing the results to
actual conditions (1) The model does not consider mass-transfer
andkinetic limitations. Inan industrial rotary kiln, the residence time
may be too short for the species to come in contact with each other
and for reactions to take place. Iurthermore, the residence time may
be too short for the transformation or decomposition of some
compounds. Therefore, the predicted formation of compounds may
exceed that occurring in reality. (2) The model does not take into
account in which form the elements are initially present in the
calcinedrawmeal and flue gas. In reality, the elements may be bound
in different minerals and be released at different temperatures. This
may affect the distribution of the element between the solid, liquid,
and gas phase. It may also affect the compounds formed and the
phase distribution of other elements.
Because of these limitations, the results of the equilibrium
calculations cannot be used for a quantitative comparison to
1ab|e 2. I|ue Gas Compositions Used in the Lqui|ibrium
Ca|cu|ations
oxidizing
conditions
( = 1.3) vol
mol/kg
of
clinker
reducing
conditions
( = 0.9) vol
mol/kg
of
clinker
CO 2;.00 ;.;80 CO 2;.00 ;.;80
H O ;.00 1.11 H O ;.00 1.11
O ;.00 1.11 O 0.00 0.000
CO 0.10 0.022 CO 2.00 0.44
SO 0.0; 0.011 SO 0.0; 0.011
N 4.8; 14.4; N .9; 1;.1
total 100.00 total 100.00
Iigure 3. Phase diagram for the SO CaO CaSO CaS CO CO system (on the basis of thermochemical data from Barin).
205
experimental data but only to determine which species are likely
to be formed under di erent gas atmospheres.
-RJPYIRGI SJ 6IHYGMRK 'SRHMXMSRW To investigate the effect
of changing the atmosphere from oxidizing to reducing, the
calcined raw meal is exposed to (a) an atmosphere with ; vol
oxygen and (b) an atmosphere with 0 vol oxygen and 2 vol
CO. The exact composition of calcined raw meal and flue gas is
given in Tables 1 and 2.
The equilibria for oxidizing and reducing conditions are shown
in Iigure 4. Only the main sulfur-containing compounds are
included for clarity. The gure shows the sulfur distribution bet-
ween CaSO , CaS, and SO at various temperatures and under
either oxidizing or reducing atmosphere. It is noted that no alkali
sulfates are formed in the studied case, because the formation of
alkali chlorides is thermodynamically favored.
The most important observation from Iigure 4 is that the
decomposition of CaSO to CaO and SO takes place at lower tem-
peratures when reducing conditions are prevailing. Under oxidizing
conditions, CaSO is stable until approximately 1100 C. At
temperatures between 1100 and 13;0 C, CaSO is gradually
decomposed, forming CaO and SO . Above 13;0 C, CaSO is
not thermodynamically stable and all sulfur is released as SO . The
situation is very di erent under reducing conditions, where CaSO
is observed to be partly reductively decomposed already at tem-
peratures below 800 C. The reductive decomposition continues
until approximately 1300 C. At higher temperatures, all sulfur is
present as SO . The reductive decomposition of CaSO takes place
because sulfates can be used as an oxygen donor under reducing
conditions. It should also be noted that a fraction of the sulfur is
present as CaS under reducing conditions. However, CaS is only
stable at temperatures below 8;0 C, and the sulfur is released as
SO at higher temperatures. This is in correspondence with the
CaSO CaS CaO SO O CO phase diagram in Iigure 3.
Through the equilibrium calculations, it was also found that the
phase distribution of Cl-containing compounds, such as HCl, NaCl,
and KCl, is not a ected by the change in the gas atmosphere. It is
identical whether the atmosphere is oxidizing or reducing. This
result was also expected because Cl is known to be much more
sensitive to the temperature than to the gas atmosphere.
The sulfur retention in the calcined cement raw materials will
depend upon the speci c degree of reducing atmosphere. The
sulfur retention under di erent atmospheres and in the tempera-
ture interval of 00 1100 C is shown in Iigure ;. This
temperature interval is representative for the conditions in the
material inlet end of cement rotary kilns, where fully or partly
calcined raw materials enter the rotary kiln with a temperature of
approximately 900 C and where the kiln ue gas leaves the
rotary kiln with a temperature of approximately 1000 1100 C.
It is observed that all sulfur is retained in the calcined raw meal
under oxidizing conditions with ; 21 O . These two curves
are identical and at 100 sulfur retention in the temperature
interval. Also, under inert conditions with 0 O , the sulfur
retention is observed to be 100, except above 10;0 C, where a
minor sulfur release is observed.
The sulfur release increases signi cantly when reducing con-
ditions prevail. The sulfur release does not increase with an
increasing concentration of the reducing agent CO. Instead, the
situation can be divided into mildly reducing conditions with
1 3 CO and strongly reducing conditions with 4 10 CO.
The sulfur release is observed to be most signi cant under mildly
reducing conditions, where it accelerates at temperatures above
80 C. At 1 CO, a fraction of the sulfur is released because of
reductive decomposition of CaSO . However, most of the sulfur is
still retained in the calcined raw meal. The thermodynamic calcula-
tions suggest that the sulfur is stabilized as K SO and K Ca (SO )
at 1 CO. The 2 CO is observed to lead to a high sulfur release,
reaching ;0 at 9;0 C. This is because the compound K Ca -
(SO ) is not stable under this reducing atmosphere, but at the same
time, the reducing atmosphere is not strong enough to stabilize
sulfur as CaS. Thus, phase equilibria are shifted toward free SO .
Iigure s. Sulfur retention in calcined raw meal as a function of the
temperature and gas atmosphere. Note the cutted y axis.
Iigure 4. Iquilibrium calculations for S distribution in calcined raw
meal. The gas atmosphere is either oxidizing with ; vol O ( = 1.3) or
reducing with 2 vol CO ( = 0.9).
206
The sulfur release is lower at 3 CO because this stronger reducing
atmosphere favors the formation of CaS. However, the thermal
stability of CaS decreases as the temperature increases, therefore, at
10;0 C, the sulfur release reaches the same level as for 2 CO.
At strongly reducing conditions, 4 10CO, the sulfur release
is observed to be lower than under mild reducing conditions.
This is again because sulfur is retained in the calcined rawmeal as
CaS. However, as the temperature increases and approaches
1100 C, the sulfur release increases rapidly because of the
thermal instability of CaS. These observations are in good
correspondence with the phase diagram shown in Iigure 3,
which indicates that sulfur may be stabilized as either CaSO at
a low reducing potential or CaS at a high reducing potential.
6IPEXMZI 7XEFMPMX] SJ 7YPJEXIW XS[EVH6IHYGMRK'SRHMXMSRW
Sulfur introduced to the kiln system with the fully or partly
calcined raw materials will mainly be in the form of sulfates,
CaSO , Na SO , and K SO . The distribution is dependent upon
parameters such as concentrations of these individual elements,
level of chlorine available to react with the alkali metals,
temperature, etc. Because it is clear that the stability of sulfates
is affected by the gas atmosphere, the relative stability of the
sulfates has been studied through equilibrium calculations.
On the basis of these equilibrium studies of the three dominating
sulfates present in the kiln system, it is clear that reducing conditions
generally shift the equilibriumtoward more SO in the gas phase. The
sulfates are used as an oxygen donor in the oxidation of CO to CO .
An interesting observation was that the relative stability of the
sulfates was Na SO > K SO CaSO , an observation that was
later con rmed experimentally. CaSO is thus less stable and more
critical thanthe alkali sulfates. This observation may indicate that it is
desirable to bind as much sulfur as possible as alkali sulfates, to keep
the sulfur evaporation in the kiln system as low as possible.
v)<4)6-1)28%0 7)'8-32
The release of sulfur from cement raw materials as a function of the
gas atmosphere was quanti ed in the temperature interval of 900
1100 C. Di erent gas atmospheres was tested, but most of the experiments
were conducted with ; vol O in N to represent an oxidizing atmosphere
and 2 vol CO in N to represent a reducing atmosphere. The temperature
interval and gas atmospheres resemble realistic temperatures and atmospheres
in the material inlet end of cement rotary kilns. The usedcement rawmaterials
Iigure o. Ixperimental setup.
Iigure . Sketch of the tube furnace reactor with size speci cations of the reactor tubes (| , outer diameter, | , inner diameter, and |, length).
1ab|e 3. Composition o Ca|cined Raw Mea| Used or the
Lxperiments
component amount (wt )
CaO
SiO 20
Al O ;
Ie O 3
CaSO ;
207
were based on a synthetic mixture of pure chemicals of CaO, SiO , Ie O ,
Al O , anda sulfate, suchas CaSO , Na SO , or K SO . These chemicals were
mixed in the right proportions to simulate a simpli ed calcined rawmeal, with
a well-knowncontent of sulfur. The compounds are also representative for the
form that the elements have when they enter the rotary kiln.
0EFSVEXSV]7GEPI 8YFI 6IEGXSV 7IXYT A laboratory-scale tube
reactor setup was used for the experiments (see Iigure ). It consists of a
gas mixing panel to mix the gases in the desired concentration and flow,
the electrically heated reactor in which the sample is placed, filters to
remove particles and moisture, a pump, and the analysers that measure
the gas concentration. Measured gas concentrations and temperatures
are collected in a computer.
The reactor depicted in Iigure is a two-zone electrically heated tube
furnace in which a horizontal alumina tube is mounted, having water-
cooled anges at both ends. The tube furnace has a maximum operating
temperature of 11;0 C. The internal diameter of the inner tube is 4
mm, and it has a heated length of 900 mm. The sample is placed in a
platinum/rhodium holder (9; Pt). The dimensions of the sample holder
are H W | = 2 3 10 cm, corresponding to a volume of
approximately 0 mI. During an experiment, the sample holder is placed in
the center of the tube furnace and the tube furnace is sealed with a stainless-
steel plate, which contains an opening for gas inlet. The gas composition
exiting the reactor is monitored on the computer. The sample will be heated
from room temperature to the desired temperature by conduction,
convection, and radiation from the tube walls and gas.
To test whether the reactor setup are producing stable and su cient
gas temperatures, center-line temperature pro les in the tube reactor
were established. It is concluded that, at the position of the sample
holder, the measured temperature is within ; C of the reactor set
point temperature. The mass ow controllers were calibrated prior to
the experiments. The actual gas ows were found to be within 4 of
the set point values on the mass ow controllers. The uncertainty on the
O , CO, and CO analysers is 2 of the total span range of each
species, while it is ; on the SO measurement.
1IXLSH The composition of used calcined cement raw meal is
shown in Table 3. Only the main components as well as a sulfur-
containing component are included. The sulfur-containing component
is CaSO , which is believed to be the dominant sulfur-containing
component in calcined raw meal entering cement rotary kilns, because
of the excessive amounts of calcium relative to alkaline metals. It should
be noted that the sulfur content in the calcined rawmeal is relatively high
(1.1 wt ) but not unrealistic. Prior to the experiments, the synthetic
calcined raw meal was stored under an inert atmosphere to prevent CaO
from reacting with moisture or atmospheric CO .
The following procedure is followed during each experiment A total
of ; g of the calcined raw meal sample is placed in a well-distributed layer
in the sample container. The sample amount is chosen to have a suitable
amount of sulfur in the experiments to detect sulfur releases in the parts
per million (ppm) range of 0 1000 ppm. The sample container is then
placed in the tube furnace at the desired temperature. A thermocouple is
placed in the center of the sample container to continuously log the
temperature. The tube furnace is sealed, and the sample is heated for 10
min in an oxidizing atmosphere of ; O in N . Practically no SO
release was observed under the 10 min heating time. After 10 min, the
gas atmosphere can be changed to the desired atmosphere, whereby the
experiment begins. The gas ow is kept at 10 NI/min at all times. The
gas composition is logged with 1 s intervals for subsequent evaluation of
the sulfur release. In all experiments, the sulfur release was measured by
integrating the SO signal over time. The experiments always lasted 20
min, not including the 10 min heating time. The SO concentration was
measured in parts per million by volume (ppmV), while O , CO , and
CO concentrations were measured in volume percentages (vol ).
In addition to the xed gas atmosphere, experiments were also
conducted with an alternating gas atmosphere between oxidizing and
reducing. In these experiments, 2 min intervals were used 2 min with
reducing conditions (2 vol of the reducing agent in nitrogen),
followed by 2 min with oxidizing conditions (; vol O in N ). This
procedure was repeated over 20 min, which gave time for ve reducing
intervals and ve oxidizing intervals.
%WWYQTXMSRW ERH 9RGIVXEMRXMIW This section discusses impor-
tant assumptions and uncertainties that may influence the obtained
results. (1) It has been experimentally determined that the sample is
Iigure 8. Concentrations of SO , O , CO, and CO during an experiment with calcined rawmeal under alternating oxidizing and reducing conditions at
1100 C and gas ow = 10 NI/min.
208
heated for approximately 2 ; min in the tube furnace before the
sample reaches the same temperature as the tube furnace. The exact
heating time depends upon the sample size and temperature to be
reached, a ; g sample, which is the sample size used in the experiments,
uses approximately 2 min to reach 900 C. To ensure sufficient time for
sample heat up, the sample will stay 10 min in the furnace under
oxidizing conditions before an experiment is started. (2) The gas
passing through the reactor will be pure N /O /CO/H /CH in
different compositions. Under industrial-scale conditions, the gas will
also contain H O, CO , NO , SO , and various volatile species, such as
alkaline metals or trace metals. It is assumed that the simplified gas
composition used in the experiments contains the most important
components and that the influence of the other components on the
results can be neglected. (3) It is assumed that all sulfur released from
the sample will form SO before leaving the reactor. This assumption
has been confirmed by two mass balances over two different calcined
rawmeal samples. After an experiment, the residual sulfur content of the
sample was determined by chemical analysis and compared to the sulfur
amount that was found in the flue gas as SO . In the first case, the sulfur
mass balance was found to be 93, and in the second case, 9; of the
sulfur was found in either the sample or the flue gas as SO . Considering
the uncertainties with sample composition, precision of analysers and
mass flow controllers, as well as accuracy of the chemical analysis
procedure, it is fairly safe to assume that nearly all released sulfur will be
in the form of SO . This assumption is also confirmed by equilibrium
calculations with the thermodynamic software IactSage, which calcu-
lates that nearly all sulfur is present as SO under the studied conditions.
(4) It is known from earlier experiments that the physical shape of the
sample holder and sample mass will cause diffusion limitation of gas
components, such as CO, to the sample. However, because all experi-
ments are made with similar sample masses and procedures, it is
assumed that results will be comparable. (;) The flue gas composition
will not change significantly in the distance between the reactor and the
analysers. To minimize this uncertainty and prevent, e.g., SO absorp-
tion in filters and condensers, these parts were regularly cleaned or
replaced to minimize the presence of compounds able to absorb SO . A
gas with a known content of SO was also regularly sent through the
system to test if any SO was absorbed before reaching the analysers.
() The calcined raw meal to be used in the experiments is representa-
tive for calcined raw meal from an industrial plant. However, an
industrial calcined raw meal will contain many minor components,
which could have an effect on results.
v6)79087 %2( (-7'977-32
Iigure 8 shows an example of the measured concentrations
during an experiment with alternating oxidizing and reducing
conditions at 1100 C. The rst 10 min is the heating period
under oxidizing conditions, initially atmospheric air with 21
vol O , followed by 4.; vol O in N when the tube reactor
is sealed. During the heating under oxidizing conditions, no
SO is observed. After 10 min, the gas concentration is changed
to 2 vol CO in N . Because the change is rapid, some oxygen
will be left in the reactor and will react with CO, forming CO .
Thus, the CO concentration is observed to increase to
approximately 1.; vol . After the last O in the reactor is
consumed, the CO concentration increases to approximately
1.8 vol . At the same time, a dramatic increase in the SO
concentration is observed from 0 to 800 ppm. The formed CO
is also observed at a level of around 0.2 vol . The SO level
drops relatively fast. When the atmosphere is again changed to
oxidizing (4.; vol O ), a new, smaller SO peak is observed.
The sulfur release cycle is observed to be repeated for each of
the following reducing/oxidizing intervals, with a slight de-
crease in SO each time, because the sulfur content in the
sample gradually decreases. The experiment is stopped after
30 min when the fth oxidizing interval is nished and the SO
concentration has dropped to values close to 0 ppm. However,
it should be noted that more sulfur would be released from the
sample if the cycles were repeated. This was tried out, and it can
be concluded that the measured SO level gradually decreases
with time but that it takes more than 2 h before all sulfur has
been released from the sample.
It is interesting to note from Iigure 8 that the rst SO peak
during the reductive decomposition of CaSO only reaches 80
ppm, while the next SO peak reaches 890 ppm. The rst SO
peak is also wider than subsequent SO peaks. This tendency was
observed in all experiments with alternating reducing and
oxidizing conditions. The same tendency has also been reported
by other researchers. The reductive decomposition of
CaSO is initiated by an induction period, which has been
explained by the slow desorption of SO and slow rate of
nucleation of CaO. Ghardashkhani et al. reported that shifting
from reducing to oxidizing and back to reducing conditions
during reductive decomposition of CaSO did not give rise to a
new induction period. This is in agreement with the experi-
mental observations, where the second SO peak reaches larger
values than the rst SO peak.
The characteristic SO peaks observed from Iigure 8 during
alternating reducing and oxidizing conditions can be explained by
the existence of phase equilibria between di erent sulfur species.
Under oxidizing conditions, sulfur is stabilized in the form of
CaSO , and under reducing conditions, sulfur is stabilized as CaS.
When the gas atmosphere changes, sulfur may be released to the
gas phase as SO , as illustrated in Iigure 2.
6IPIEWI 5YERXMJMGEXMSR The release of sulfur during the
experiments is calculated by integration of the gas concentration
of SO over the relevant time interval
n
sulfur found

>
t
0
y
SO2
dt 10

|V
P1
10
where y is in ppmV and V is the gas flow in I/s. It should be
noted that, because of diffusion limitation of gas components,
such as CO, only experiments made with similar sample masses
can be compared. The sum of sulfur found as SO is divided by
Iigure 9. I ect of the gas atmosphere on the sulfur release from
calcined raw meal. Sulfur release was evaluated after 20 min of
reaction time.
209
the total sulfur amount in the sample
S release
n
sulfur found
n
sulfur in sample
100 11
)JJIGX SJ XLI 6IHYGMRK %XQSWTLIVI [MXL '3 The sulfur
release from calcined raw meal has been studied at the tempera-
tures of 900, 1000, and 1100 C. Two different atmospheres with
; O and 2CO (pure nitrogen is the balance in both cases) as
well as alternating reducing and oxidizing conditions have been
studied. The results are shown in Iigure 9. Under oxidizing
conditions with ; oxygen, the S release is approximately 0 in
the studied temperature interval. This result has been found to be
valid at all oxygen concentrations from 1 to 21, indicating that
CaSO is generally stable under oxidizing conditions and tem-
peratures below 1100 C. This finding is in agreement with the
thermodynamic equilibrium calculations, which did not show
any decomposition of CaSO below 11;0 C.
Under reducing conditions with 2 CO, the sulfur release was
still approximately 0 at 900 C but increased to 1at 1000 C
and 19 at 1100 C. This signi cant increase in sulfur release
under reducing conditions may be explained by the formation of
CaS from CaSO , in which CaO and free SO are reaction
intermediates. Iach experiment was repeated 3 times, and the
standard deviation were found to be 2.
Under shifting oxidizing and reducing intervals with 2 min of
2 CO, followed by 2 min of ;O continuously repeated 2 ;
times, the sulfur release is observed to be signi cantly higher than
under constant reducing or oxidizing conditions 14 at 900 C,
40 at 1000 C, and 48 at 1100 C. Iach experiment has been
repeated 3 times, and the standard deviation was found to be
3. This strong e ect of alternating oxidizing and reducing
conditions is in correspondence with the sulfur transformation
cycle shown in Iigure 2. Under oxidizing conditions, sulfur is
stabilized as CaSO , and under reducing conditions, sulfur is
stabilized as CaS. Ivery shift in atmosphere leads to the formation
of CaO and free SO . The formation of free SO is particularly large
at higher temperatures, whereas the lowest temperature of 900 C
can limit the sulfur release even under alternating oxidizing and
reducing conditions.
The results are in correspondence with previous work, where
sulfur release and capture has been studied under alternating
oxidizing and reducing conditions. Iyngfelt and Ieckner re-
ported a decrease in the sulfur capture ability of limestone at
temperatures above 880 C and reducing conditions, because of
reductive decomposition of CaSO . Hansen et al. found that the
competition between sulfur capture and sulfur release under
alternating oxidizing and reducing conditions was shifted toward
more sulfur release when the temperature increased. Tarelho
et al. found a maximal sulfur capture e ciency of limestone at
around 82; C. Higher temperatures reduced the sulfur capture
e ciency, particularly under reducing conditions.
)JJIGX SJ 3XLIV 6IHYGMRK %KIRXW Most investigations of
reductive decomposition of CaSO reported in the literature use
either CO or H as the reducing agent. Because combustion of
alternative fuels in cement kilns may lead to the formation of
significant amounts of CO, H , and hydrocarbons, such as CH ,
under substoichiometric conditions, these three reducing agents
have been investigated with respect to sulfur release from
calcined raw meal in the temperature interval of 900 1100 C.
The results are shown in Iigure 10. The experimental conditions
were shifting oxidizing and reducing intervals with 2 min of the
reducing agent (2of either CO, H , or CH ), followed by 2 min of
;O . The 2 min intervals were continuously repeated 2 ; times,
thus 20 min of total reaction time. All experiments were repeated 3
times with a standard deviation in the order of 3. Under these
reaction conditions, the sulfur release when COwas used was 1;, 40,
and 48 at 900, 1000, and 1100 C, respectively. When CH was
used, the sulfur release was , 40, and 1 at 900, 1000, and
1100 C, respectively. Iinally, when H was used, the sulfur release
was 8, 2, and 20 at 900, 1000, and 1100 C, respectively.
While the results with CO are observed to lead to increased
sulfur release as a function of the temperature, the results with H
and CH are more surprising. The sulfur release increases from
900 to 1000 C but then decreases again at 1100 C. This was
unexpected because it was assumed that the sulfur release would
increase with the temperature, regardless of the reducing agent. It
was also expected that H would lead to a higher sulfur release
than CO and CH would lead to a much lower sulfur release than
CO, according to Hansenvs ndings. Hansen found that the rate
of reductive decomposition of CaSO increased when H was
used instead of CO and that CH gave no reductive decomposi-
tion at all. However, the reaction conditions used in Hansenvs
experiments were quite di erent from the ones used in this
investigation, because he used relatively short time intervals of 30
s/30 s and only studied the e ect of the reducing agent at 8;0 C.
In this experiment, longer time intervals of 2 min/2 min were
used and the temperature interval was 900 1100 C. This may
explain why CH is found to be of comparable e ciency as H
and CO in this investigation, while Hansen reported it to be less
e cient because of slower reaction kinetics.
With regard the lower sulfur release at 1100 C relative to
1000 C when H and CH were used as reducing agents, similar
results have been reported by Kamphuis et al. At temperatures
above 9;0 C, a fraction of SO was observed to be converted to
H S, which was not detected by the analysis system, when the
ue gas contained H . On the basis of a series of tests with
known SO concentrations in a H /N gas mixture passing
Iigure 10. I ect of the reducing agent on sulfur release from calcined
raw meal under alternating oxidizing and reducing conditions. Sulfur
release was evaluated after 20 min of reaction time.
210
through the reactor without a sample, Kamphuis et al. established
a correction factor to correlate for the formed H S. It is likely that
the same phenomenon took place in the present experiments
with H and CH because H S can be formed fromboth reducing
agents. This will explain why the found sulfur release is lower at
1100 C relative to 1000 C. Prior to the experiments, thermo-
dynamic equilibrium calculations had been made for the calcined
raw meal exposed to the reducing agents H and CH . These
thermodynamic calculations indicated that practically all released
sulfur would be present as SO under the experimental condi-
tions. However, the thermodynamic calculations may be mis-
leading because they do not consider reaction kinetics. H S may
be formed as an intermediate species, which leaves the reactor
before the formation of SO occurred. A sulfur mass balance
should be performed to determine whether some sulfur has
escaped from the system as H S or not. Because a sulfur mass
balance has not been conducted for the experiments with H and
CH , it is not directly possible to quantify the e ect on sulfur
release as a function of the reducing agent. However, the results
do indicate that all three reducing agents, CO, CH , and H , have
an e ect on the sulfur release in the studied temperature interval.
v'32'097-32
The release of sulfur from cement raw materials under both
oxidizing and reducing conditions was investigated. The inves-
tigations included thermodynamic equilibrium calculations in
the temperature interval of 800 1;00 C and experiments in a
tube furnace reactor in the temperature interval of 900 1100 C.
The investigated conditions resemble actual conditions in the
material inlet end of cement rotary kilns, where local reducing
conditions may occur during combustion of solid, alternative fuel
particles and where solid/gas temperatures are usually between
900 and 1100 C. The following results were obtained (1) The
thermodynamic equilibrium calculations showed that the stabi-
lity of sulfates present in calcined cement raw meal is strongly
a ected by the gas atmosphere. Ispecially, CaSO tends to
decompose at much lower temperatures under a reducing atmo-
sphere. K SO and Na SO are also a ected by a reducing
atmosphere but not as signi cant as CaSO . (2) Ixperimentally,
a clear tendency for increasing sulfur release was observed when
the gas atmosphere shifted from oxidizing to reducing. No sulfur
release was observed under oxidizing conditions (; vol O ),
whereas the sulfur release increased to 19 under reducing
conditions (2 vol CO) at 1100 C. The sulfur release was
particularly signi cant if the calcined raw meal was exposed to
alternating oxidizing and reducing conditions, which led to a
sulfur release of 48 at 1100 C. (3) The sulfur release from
calcined raw meal was observed to be dependent upon the
temperature. Under alternating oxidizing and reducing condi-
tions, the sulfur release was observed to be 14 at 900 C and
48 at 1100 C. (4) Sulfur release from calcined raw meal was
shown to take place in the presence of any of the reducing agents
CO, H , and CH . However, it was not possible to directly
quantify the sulfur release as a function of the reducing agent,
because part of the released sulfur from the experiments with H
and CH presumably formed H S, which could not be detected
by the analysis system.
v%98,36 -2*361%8-32
'SVVIWTSRHMRK %YXLSV
*Telephone 4;-4;2;2831. I-mail arnkt.dtu.dk.
v%'/23;0)(+1)28
The work described in this paper is part of a research platform
on future cement technology nanced by The Danish National
Advanced Technology Ioundation, Technical University of
Denmark (DTU) and IISmidth A/S.
v6)*)6)2')7
(1) Verein Deutscher Zementwerke (VDZ). Act|r|ty Pcport 200
2009, 2009, http//www.vdz-online.de (accessed on April 20, 2011).
(2) Cembureau. Act|r|ty Pcport 2008, http//www.cembureau.be
(accessed on April 20, 2011).
(3) BP. Stot|st|co| Pcr|ci oj Wor|d Lncrgy, June 2009, http//www.
bp.com (accessed on April 20, 2011).
(4) Cembureau. A|tcrnot|rc |uc|s |n ccmcnt Monujocturc, 199,
http//www.cembureau.be (accessed on April 20, 2011).
(;) Iyngfelt, A., Ieckner, B. c|cm. Lng. j. 1989, +0, ;9z9.
() Hansen, J. P. SO emissions from cement production. Ph.D.
Thesis, Department of Chemical Ingineering, Technical University of
Denmark (DTU), Iyngby, Denmark, 2003, ISBN 8-90142-9-9.
() Choi, G.-S., Glasser, I. P. ccm. concr. Pcs. 1988, 18, 3z34.
(8) Iarag, I. M., Kamel, H. M. ZkG |nt. 1994, 10, ;8z;90.
(9) Rosholm, P., Mortensen, A. H., Hintsteiner, I. A. ZkG|nt. 1998,
;1 (2/4), 19.
(10) Twomey, C., Birkinshaw, C., Breen, S. j. c|cm. 1cc|no|.
8|otcc|no|. 2004, 9, 48z490.
(11) Klauss, J. ZkG |nt. 2000, ;3 (3), 132z144.
(12) Sylla, H.-M. ZkG |nt. 194, 10, 499z;08.
(13) Iyngfelt, A., Ieckner, B. c|cm. Lng. Sc|. 1989, ++ (2), 20z213.
(14) Iyngfelt, A., Ieckner, B. c|cm. Lng. Sc|. 1999, ;+, ;;3z;;84.
(1;) Dam-Johansen, K., Hansen, P. I. B., stergaard, K. c|cm. Lng.
Sc|. 1991, +o (3), 84z8;3.
(1) Hansen, P. I. B., Dam-Johansen, K., Bank, I. H., stergaard, K.
|roc. |nt. conj. ||u|d. 8cd com|ust. 1991, 1, 3z82.
(1) Hansen, P. I. B, Dam-Johansen, K., stergaard, K. c|cm. Lng.
Sc|. 1993, +8 (), 132;z1341.
(18) Tarelho, I. A. C., Matos, M. A. A., Pereira, I. J. M. A. |uc|
|roccss. 1cc|no|. 200s, 8o, 138;z1401.
(19) c|cm|stry oj ccmcnt ond concrctc, 4th ed., Hewlett, P. C., Id.,
John Wiley and Sons, Inc. New York, 1998, ISBN 0-340-;;89-.
(20) Tian, H., Guo, Q. |nd. Lng. c|cm. Pcs. 2009, +8, ;24z;32.
(21) Barin, I. 1|crmoc|cm|co| |oto oj |urc Su|stonccs, 3rd ed., VCH
Weinheim, Germany, 199;, ISBN 3-;2-284;-0.
(22) Diaz-Bossio, I. M., Squier, S. I., Pulsifer, A. H. c|cm. Lng. Sc|.
198s, +0 (3), 319z324.
(23) Wheelock, T. D., Boylan, D. R. |nd. Lng. c|cm. 19o0, ;2 (3),
21;z218.
(24) Turkdogan, I. T., Vinters, J. V. 1rons. |nst. M|n. Mcto||., Scct. c
19o, 8;, 113z123.
(2;) Iocher, G., Klein, H. ccmcnt |nt. 2009, (4), 4z;.
(2) Ghardashkhani, S., Ijungstrom, I., Iindquist, O. |roc. |nt. conj.
||u|d. 8cd com|ust. 1989, 11z1;.
(2) Kamphuis, B., Potma, A. W., Prins, W., Van Swaaij, W. P. M.
c|cm. Lng. Sc|. 1993, +8 (1), 10;z11.
211
Published: August 15, 2011
r2011 American Chemical Society 3917 dx.doi.org/10.1021/ef200963v
|
Energy Fuels 2011, 25, 39173924
ARTICLE
pubs.acs.org/EF
Sulfur Release from Cement Raw Materials during Solid
Fuel Combustion
Anders R. Nielsen,
,,
* Morten B. Larsen,

Peter Glarborg,

and Kim Dam-Johansen

FLSmidth A/S, Vigerslev Alle 77, DK-2500 Valby, Denmark

Department of Chemical and Biochemical Engineering, CHEC Research Centre, Technical University of Denmark (DTU),
DK-2800 Lyngby, Denmark
ABSTRACT: During combustion of solid fuels in the material inlet end of cement rotary kilns, local reducing conditions can occur
and cause decomposition of sulfates from cement raw materials. Decomposition of sulfates is problematic because it increases the
gas-phase SO
2
concentration, which may cause deposit formation in the kiln system. SO
2
release from cement raw materials during
combustion of solid fuels has been studied experimentally in a high temperature rotary drum. The fuels were tire rubber, pine wood,
petcoke, sewage sludge, and polypropylene. The SO
2
release from the raw materials was observed to increase when (a) the inlet
oxygen concentration decreased, (b) the temperature increased, and (c) when the total surface area of the fuel particles increased.
The type of fuel also had a signicant eect on the SO
2
release. The sequence of SO
2
release for fuel particles in the size interval 12
mm was found to be tire rubber granulate > sewage sludge > pine wood sawdust > petcoke > polypropylene akes. The SO
2
release
was generally observed to increase when formation of local reducing conditions near the raw material bed increased, as indicated by
elevated levels of CO during the fuel devolatilization.
INTRODUCTION
Fuel consumption accounts for a signicant fraction of the
total cement clinker production costs.
1
Traditionally, cement
production has mainly depended on the fossil fuels coal, oil, and
natural gas. Due to competition in the cement market, rising
fossil fuel prices, and environmental concerns, cement producers
have increased the utilization of alternative fuels as a substitute
for fossil fuels in order to achieve the most economic fuel mix. In
this context, alternative fuels cover all nonfossil fuels and waste
from other industries. Popular alternative fuels in the cement
industry are tire-derived fuels, biomass residues, and dierent
commercial and industrial wastes.
It is attractive to utilize coarse, solid alternative fuel particles in
the material inlet end of cement rotary kilns in order to save the
expense of shredding of the fuels to smaller particles and to
increase fuel exibility of the system. High temperatures in the
rotary kiln and material retention times of several minutes
provide good conditions for fuel burnout. Figure 1 show how
coarse, solid alternative fuels can be introduced into the material
inlet end of a cement rotary kiln.
Sulfur is introduced to cement rotary kilns as inorganic bound
sulfur in the preheated and calcined cement rawmaterials or with
the fuels (see Figure 1). A detailed description of sulfur species
and the main sulfur reactions in rotary kilns and calciners is
described elsewhere.
2,3
Sulfur present in the calcined raw material is mainly bound as
CaSO
4
or as alkali sulfates. The calcined raw material is trans-
ported by the rotational movement and slight inclination of the
rotary kiln toward the hot burning zone, where ame tempera-
tures are in the range of 2000 C and material temperatures are
around 1450 C. The clinker reactions take place during the
gradual heating of the calcined raw material as it is transported
through the rotary kiln. During the gradual heating of the calcined
raw material, a sulfate melt is formed that facilitates formation of
alkali sulfates, Na
2
SO
4
and K
2
SO
4
, which are more thermody-
namically stable than CaSO
4
.
3
CaSO
4
will decompose at the high
temperatures present in the burning zone and form SO
2
or alkali
sulfates.
The main challenge with respect to alternative fuel utilization
in the material inlet end of rotary kilns is that solid fuel particles
will be in physical contact with the calcined raw material: During
the fuel devolatilization, reducing agents such as CO, H
2
, and/or
hydrocarbons are formed. These reducing species may react with
elements in the calcined raw material before they are oxidized to
their ultimate combustion products, H
2
O and CO
2
. In addition,
if the fuel particles are fully or partly covered by calcined raw
material, mass transfer of oxygen to the fuel char will be hindered.
Substoichiometric amounts of oxygen will lead to incomplete
oxidation of the fuel char, forming reducing agents in the form of
CO, H
2
, and/or hydrocarbons. The process stability of the kiln
system may be aected by increased release of sulfur from the
calcined raw material, mainly by reductive decomposition of
CaSO
4
:
CaSO
4
s COg T CaOs CO
2
g SO
2
g
1
Increasing amounts of SO
2
in the cement kiln system is
problematic because SO
2
promotes formation of sulfo-
spurrite (2(2CaO
3
SiO
2
)
3
CaSO
4
) and calcium sulfoalumi-
nate (3CaO
3
3Al
2
O
3
3
CaSO
4
), some of the principal constituents
of deposit buildups found in rotary kilns and kiln riser ducts.
4
Received: July 1, 2011
Revised: August 15, 2011
212
3918 dx.doi.org/10.1021/ef200963v |Energy Fuels 2011, 25, 39173924
Energy & Fuels
ARTICLE
These sulfur-containing deposits can lead to blockages that need
to be removed, sometimes by a temporary plant shutdown.
Relatively little information is available in the literature about
SO
2
release from cement raw materials during utilization of solid
fuels. However, adsorption and release of SO
2
for the SO
2

CaOCaSO
4
CaSCOCO
2
system under uidized bed con-
ditions has been studied extensively during the last decades.
511
Even though these investigations are not directly applicable to
describe the conditions in the material inlet end of cement rotary
kilns, they do include relevant descriptions of mechanisms for
SO
2
release and capture by limestone at temperatures up to
1200 C and under oxidizing as well as reducing atmospheres.
Hansen et al.
9,10
studied phase equilibria for the SO
2
CaO
CaSO
4
CaSCOCO
2
system. They performed experiments
in an electrically heated laboratory scale xed bed reactor
developed to simulate the changing oxidizing and reducing
conditions similar to the conditions particles will experience in
a uidized bed reactor. They found that any transformation
between CaSO
4
and CaS takes place via CaO. This competition
between sulfur capture and sulfur release depends on parameters
such as partial pressures of SO
2
, CO, and CO
2
(see the equilibrium
Figure 1. Illustrationof alternative fuel utilizationinthe material inlet endof a cement rotary kiln. The calciner andtwolowest preheater cyclones are alsoshown.
Figure 2. Phase diagramfor the SO
2
CaOCaSO
4
CaSCOCO
2
system (based on thermochemical data from Barin).
12
213
3919 dx.doi.org/10.1021/ef200963v |Energy Fuels 2011, 25, 39173924
Energy & Fuels
ARTICLE
phase diagram in Figure 2). Temperature is a very important
parameter: An increase in temperature in the interval 800
1200 C will shift the equilibrium toward CaO and release more
SO
2
.
9
The aim of this work is to obtain quantitative data on the
release of SO
2
from calcined cement raw materials under
conditions that resemble those in the material inlet end of rotary
kilns. The purpose is to document the impact on SO
2
release
during utilization of solid alternative fuels, in the material inlet
end of cement rotary kilns.
EXPERIMENTAL SECTION
The purpose of the experiments is to clarify how the sulfur release in
the form of SO
2
is aected by dierent solid fuels and reaction con-
ditions. The release of sulfur from cement raw materials as a function of
fuel type and fuel characteristics was quantied in the temperature
interval from 700 to 1000 C and at oxygen concentrations from 5 to
21 vol %. The temperature interval and oxygen concentrations up to
10 vol %resemble realistic material temperatures and atmospheres in the
material inlet end of cement rotary kilns. The used cement raw materials
were based on a synthetic mixture of quartz sand and 5 wt % calcium
sulfate, CaSO
4
. This represents a simplied batch of cement rawmaterial
entering the rotary kiln, with a high, but realistic, sulfur content.
Experimental Setup. The experiments were made in a high-
temperature rotary drum in a commercially available chamber furnace
(see Figure 3). The setup can reach an operating temperature of
1000 C. It is electrically heated by heating elements embedded in
two side walls and the floor. The steel drum is supported by a steel tube
on rollers that passes through the rear wall. This steel tube is connected
to a motor by means of a roller chain, in order to rotate the drum. The
part of the steel tube that is outside the chamber furnace is water cooled
in order to protect the rollers, roller chain, and gas seals against the high
temperatures.
Gas can be transported to the chamber furnace through a hole in the
roof. The furnace has been designed to achieve sucient heating of the
reactant gas before it enters the rotary drum; this was veried by
temperature measurements. The gas is transported into the rotating
steel drum due to an externally placed gas pump that pumps the gas
through the rotating drum and steel tube, after which the gas exits the
reactor for subsequent cooling and analysis.
The chamber furnace door is equipped with a window for visual
inspection, two holes for thermocouples, and a centrally placed water-
cooled tube for solid fuel addition. The solid fuels are placed in a sample
container that can be pushed into the hot rotary drum or pulled out to
the water-cooled tube.
The experimental setup is shown in Figure 4. The gas supply can be
up to 500 NL/min of air and nitrogen. The gas is transported to the
rotary drum reactor during the experiments, but may also be bypassed
during calibration of ow controllers, leakage tests, etc.
The temperature is measured in three dierent positions in the rotary
drum reactor: at the rear wall in the chamber furnace (T1), in the center
of the rotary drum (T2), and at the door just in front of the rotary drum
(T3). A pump transports the ue gas out of the rotary drum reactor.
Before passing the pump, the ue gas is cooled in a heat exchanger, and
soot particles are captured in a lter. After having passed the pump, the
ue gas is sent directly either to the stack or to the gas analyzers for
measurement of O
2
, CO, CO
2
, SO
2
, and NO. The fraction of the ue gas
that is transported to the analyzers will pass another two lters, a gas
cooler for condensation of water, and a sample gas pump before reaching
the gas analyzers.
Materials and Methods. A 2.5 kg portion of raw material in the
formof quartz sand with 5 wt %CaSO
4
is placed in the rotary drum. This
corresponds to a volumetric fill degree of 5%. The drum is rotated at
6 rpm in order to keep the raw material in a continuous rolling motion.
The raw materials are then preheated to the desired temperature, in the
interval from700 to 1000 C. Aconstant flowof 100 NL/min of gas with
the desired oxygen concentration is transported through the rotary drum
and will pass over the raw materials in the lower part of the rotary drum.
The gas is preheated to the same temperature as the raw materials in
order to obtain a uniform temperature at all positions. The temperature
and oxygen concentration are monitored continuously, and when these
have stabilized, a batch of solid fuel is rapidly added into the raw
materials in the rotary drum. SO
2
, NO, O
2
, CO, and CO
2
concentrations
are measured during the devolatilization and char combustion. The gas
composition is logged for subsequent evaluation of the sulfur release. In
all experiments, the sulfur release was measured by integrating the SO
2
signal over time:
m
sulf ur

Z
0
t
y
SO2
dt 10
6
MW
sulf ur

PV
RT
2
where y
SO
2
is the mole fraction in ppmV, MW
sulfur
is the molar weight of
sulfur in g/mol, and V is the gas flow in L/s.
Since many of the fuels contain sulfur that will also be released during
the experiments, baseline experiments have been made, where the fuels
are combusted in 2.5 kg of pure quartz sand, without CaSO
4
. The
measured SO
2
from the baseline experiments can only arise from
the fuels and not from the raw material. The amount of sulfur
released during the baseline experiments is subtracted from the sulfur
released during the actual experiment in order to get the corrected
accumulated amount of sulfur that arises from the raw material.
Figure 3. High-temperature rotary drum: (left) side view and (right) top view.
214
3920 dx.doi.org/10.1021/ef200963v |Energy Fuels 2011, 25, 39173924
Energy & Fuels
ARTICLE
All experiments were repeated at least two times to ensure the
repeatability. The sulfur release values reported in the gures are the
average values from the experiments, corrected for the sulfur contribu-
tion from the fuels.
Five dierent solid fuels have been tested: tire rubber, pine wood,
polypropylene, petcoke, and sewage sludge. In order to compare the
sulfur release between fuels, the amount of fuel is chosen to obtain the
same energy input. Fuel proximate and ultimate analyses as well as lower
heating values are shown in Table 1. The lower heating values for the
fuels are quite dierent. Thus, quite dierent fuel sample amounts have
been used in order to have the same energy input in each experiment,
regardless of the fuel. The energy input during the experiments was
chosen to be in the interval 1550 kJ per batch, corresponding to fuel
sample sizes from 0.3 to 4.2 g. These energy inputs were chosen to
obtain a suitable level of sulfur release, thereby avoiding SO
2
peaks
exceeding the analyzer range.
Several dierent fuel particle sizes and shapes have been used in the
experiments, and these are presented in Table 2.
Assumptions, Limitations, and Uncertainties. This section
discusses important assumptions, limitations, and uncertainties that may
influence the obtained results.
(1) The temperature at which the experiment is conducted is
determined as the measured temperature in the center of the
rotary drum, above the rawmaterial charge. It is assumed that the
gas temperature and raw material temperature is the same at all
positions in the rotary drum. Temperature proles have been
measured at various positions in the rotary drum, and the
temperature was found to deviate no more than (15 C.
(2) The experiments are conducted in the temperature interval from
700 to 1000 C, whereas the raw material and gas temperatures
in the material inlet end of modern industrial cement rotary kilns
are typically 900 C and 10001200 C, respectively. The
temperature interval is thus only partly representative for the
industrial conditions. However, it is not possible to conduct
experiments above 1000 C in the present experimental setup.
(3) The gas passing through the rotary drum is pure mixtures of N
2
and O
2
. During the devolatilization and char combustion, a frac-
tion of the O
2
will be consumed, and the combustion products
H
2
O, NO, SO
2
, CO, and CO
2
will be formed. In addition, some
intermediate products such as CH
4
or higher hydrocarbons,
alkali species, etc. may be formed. Under industrial conditions,
the gas will also contain the combustion products from the main
kiln burner and may have overall gas compositions dierent from
the one used in these experiments. This dierence in gas
composition is likely to aect the sulfur release to some extent.
(4) The uncertainty on the O
2
, CO, and CO
2
analyzers are (2% of
span range for each species, while it is (5% on the SO
2
measure-
ment. These uncertainties should be kept in mind when the
results are evaluated.
(5) It is assumed that all sulfur released from the sample will be
oxidized to SO
2
before leaving the reactor. This assumption is
supported by equilibrium calculations with the thermodynamics
software FactSage, which predicts that nearly all sulfur is present
as SO
2
under the studied conditions.
2
(6) The physical shape of the rotary drum reactor may cause a diu-
sion limitation of oxygen to the fuel particles, since oxygen must
be transported from the drum center line to the fully or partly
Figure 4. Sketch of the high-temperature rotary drum experimental setup.
Table 1. Fuel Analyses and Lower Heating Values (LHV) for
Solid Fuels Used in the Experiments
proximate analysis
a
ultimate analysis
a
VM FC ash C H N S LHV (MJ/kg)
tire rubber 64.6 32.6 2.8 87.4 7.1 0.3 1.2 36.9
pine wood 75.3 24.5 0.2 38.9 5.2 0.1 16.2
polypropylene 97.5 0.0 2.5 83.0 14.0 0.0 44.5
petcoke 13.4 85.1 1.5 87.3 3.7 1.5 4.7 34.0
sewage sludge 49.3 0.5 50.2 29.0 3.8 3.8 1.0 12.5
a
Units are weight percent as received.
Table 2. Approximate Shapes and Dimensions of the Fuels
Used in the Experiments
fuel shape dimensions
tire rubber granulate irregular D
p
2 mm
tire rubber cylinders cylindrical L = 12 mm, D
p
= 9 mm
pine wood cubes rectangular thickness 10 mm
pine wood sawdust needles thickness 1 mm
polypropylene akes rectangular thickness 1 mm
petcoke spherical D
p
1 mm
sewage sludge spherical D
p
0.5 mm
215
3921 dx.doi.org/10.1021/ef200963v |Energy Fuels 2011, 25, 39173924
Energy & Fuels
ARTICLE
buried fuel particles in the raw material bed. However, since all
experiments are made under similar conditions, it is assumed that
results will be comparable.
(7) The ue gas composition will not change signicantly in the
distance between the reactor and the analyzers. To minimize this
uncertainty and prevent, for example, SO
2
absorption in lters
and condensers, these parts were regularly cleaned or replaced to
minimize the presence of compounds able to absorb SO
2
. Agas with
known content of SO
2
was also regularly sent through the systemto
test if any SO
2
was absorbed before reaching the analyzers.
RESULTS AND DISCUSSION
General Observations. Figure 5 shows an example of the flue
gas composition during combustion of tire rubber granulate at an
initial oxygen concentration of 10 vol %. When the devolatiliza-
tion starts, the O
2
concentration drops to approximately 7 vol %
and the CO
2
concentration increases to nearly 4 vol %. The CO
concentration also increases shortly to around 600 ppm. The
change in flue gas composition is followed by a rapid increase in
the SO
2
concentration from 0 ppm to around 2500 ppm. The
SO
2
concentration then gradually decreases toward 0 ppm again,
as the CO/CO
2
concentrations gradually decrease.
The sulfur release from the tire rubber and raw material is
observed to begin almost immediately during the fuel heat up and
devolatilization. Baseline experiments with fuel combustion in
pure quartz sand without CaSO
4
shows that only a minor fraction
of the sulfur release arises from fuel-bound sulfur; most of the
sulfur is released from the raw materials, which is also shown in
Figure 5. The sulfur release is presumably facilitated by the
formation of local reducing conditions near the raw material
charge during the fuel devolatilization, leading to reductive
decomposition of CaSO
4
. The tendency for a rapid sulfur release
during the fuel devolatilization was observed for all experiments
regardless of the fuel.
Effect of Energy Input and Fuel Type. The sulfur release
from the raw material as a function of the energy input has been
studied for five different fuels at 900 Cand 10 vol %oxygen. The
fuels were polypropylene flakes, petcoke, pine wood sawdust, tire
rubber granulate, and sewage sludge (see Tables 1 and 2 for
details). These fuels all have relatively similar average particle
thicknesses in the interval 0.52.0 mm. The results are shown in
Figure 6. For all tested fuels it is observed that the sulfur release
from the raw materials increases as the energy input increases. It
is also observed that the sulfur release is quite different depend-
ing on the specific fuel: tire rubber granulate and sewage sludge
leads to the highest sulfur release, while pine wood, petcoke, and
polypropylene all leads to a lower sulfur release.
The reason why the sulfur release increases with increasing
energy input becomes clear when the ue gas compositions at
each individual experiment are compared with each other: The
larger the energy input, the more the oxygen concentration will
decrease during the fuel devolatilization, while both COand CO
2
concentrations increase, thereby forming a higher degree of
reducing conditions near the raw material charge.
The large dierences in sulfur release for identical energy
inputs but by dierent fuels are interesting. A detailed study of
the ue gas composition for each fuel type has not revealed any
clear trends about the degree of reducing conditions and sulfur
release. The reason is probably that the exact particle sizes,
shapes, densities, fuel combustion pathways, and kinetics of the
studied fuels are quite dierent and that this also has an eect on
sulfur release. The approximately spherical tire rubber granulate
and sewage sludge particles may for example lead to a high sulfur
release because of a relatively fast devolatilization combined with
a high degree of mixing with the rawmaterial. The needle-shaped
sawdust particles yield a lower sulfur release than tire rubber
granulate and sewage sludge, even though the sawdust also has a
fast devolatilization. An explanation could be that the sawdust
particles are less dense than the tire rubber and sewage sludge
and that the density dierence leads to a low degree of mixing of
sawdust into the raw material. The polypropylene akes have the
lowest sulfur release, which may be explained by a very fast
devolatilization, so fast that the polypropylene devolatilization
ends before it has been well mixed into the raw material. Finally,
the small petcoke particles are a slow burning fuel with a volatile
content of only 13%. This leads to a relatively small decrease in
oxygen concentration during the fuel conversion compared to
Figure 5. Concentrations of O
2
, CO
2
, CO, and SO
2
during an experi-
ment with 0.81 g of tire rubber granulate corresponding to an energy
input of 30 kJ. Conditions: 900 C, 5%ll, 6 rpm, 100 NL/min, 10% O
2
.
Figure 6. Corrected accumulated sulfur release as a function of energy
input from dierent fuels. Conditions: 900 C, 10 vol % O
2
, 5% ll, 100
NL/min, 6 rpm.
216
3922 dx.doi.org/10.1021/ef200963v |Energy Fuels 2011, 25, 39173924
Energy & Fuels
ARTICLE
the other fuels and thereby a low degree of local reducing
conditions near the raw material bed.
Effect of Temperature and Fuel Particle Size. The sulfur
release is studied as a function of the temperature and fuel
particle size, which are both important for the devolatilization
and char combustion times. Tire rubber and pine wood are used
in the investigations about particle size while the investigations
about temperature also include petcoke and polypropylene. The
tire rubber is fed as cylinders or tire rubber granulates, while the
pine wood is in the form of one pine wood particle or as sawdust.
The overall energy input is the same in each experiment. Details
about the fuels can be found in Tables 1 and 2. The results are
shown in Figure 7 with sulfur release in milligrams versus temper-
ature for the different fuel particle sizes. It is observed that the
fuel particle size is insignificant for the sulfur release at 700 and
800 C, while the influence becomes more evident at 900 Cand,
in particular, at 1000 C: At 900 and 1000 Cthe sawdust and tire
granulate lead to a higher sulfur release than for the wood particle
and tire particle.
The signicant eect of temperature on the sulfur release was
also to be expected due to the lower thermal stability of CaSO
4
at
higher temperatures (see the phase diagram in Figure 2). In
addition, comparison of the ue gas compositions during the
combustion at dierent temperatures shows that the CO con-
centration during the fuel devolatilization reaches higher values
at higher temperatures than at lower temperatures. This can be
attributed to be a faster release of volatiles from the fuel particles
at elevated temperatures, thereby increasing the tendency for
local reducing conditions in the area around the fuel particles.
Local reducing conditions will augment the sulfur release during
the fuel devolatilization.
The reason for the dierence in sulfur release for dierent fuel
particle sizes becomes clear when the ue gas composition is
studied for the individual experiments: The larger total surface
areas of the smaller fuel particles leads to a fast devolatilization
and thus a fast release of volatiles. This leads to signicantly
higher CO concentrations and thus a higher reductive decomposi-
tion of CaSO
4
during the devolatilization. An example is illustrated
inFigure 8, which shows ue gas compositions during wood particle
and sawdust combustion, respectively, at 1000 C. During wood
particle combustion, the CO peak during devolatilization reaches
180 ppm and the SO
2
peak reaches 1350 ppm. During sawdust
combustion, the CO peak is signicantly larger and reaches 27 000
ppm (2.7 vol %) which leads to a SO
2
peak well above the analyzer
detection range of 3500 ppm. The changes in O
2
concentration are
from the initial 10 to 5 vol % for the sawdust and to 7 vol % for the
pine wood particle. Similar trends were observed for the experi-
ments with one tire particle versus tire granulate.
Since the experiments with sawdust and tire granulate at
1000 C led to SO
2
peaks well above the SO
2
analyzer range,
it has been necessary to estimate the total sulfur release from
these two experiments. This induces some uncertainty on the
exact amount of released sulfur in these two cases at 1000 C.
Figure 7. Corrected accumulated sulfur release from raw materials as a
function of temperature and fuel particle size. Conditions: 5%ll degree,
10 vol % O
2
, 100 NL/min, 6 rpm. Energy input in each experiment:
30 kJ. Sulfur release for tire granulate and sawdust at 1000 C has been
estimated because the SO
2
concentration exceeded the analyzer range
for these two specic experiments.
Figure 8. Flue gas composition during combustion of a pine wood particle
with dimensions 30 15 10 mmand pine wood sawdust with particle size
<1mm. Conditions: 10vol %O
2
, 5%ll degree, 1000C, 100NL/min, 6rpm.
Figure 9. Corrected accumulated sulfur release versus the total surface
area of pine wood and tire rubber fuel particles. Conditions: 900 C,
10 vol % O
2
, 5% ll degree, 6 rpm, 100 NL/min.
217
3923 dx.doi.org/10.1021/ef200963v |Energy Fuels 2011, 25, 39173924
Energy & Fuels
ARTICLE
Figure 9 shows the sulfur release versus the total surface area of
the tire rubber and pine wood fuel particles at 900 C. The sulfur
release is observed to increase approximately linearly as a
function of surface area for tire rubber. It should be noted that
the fuel mass, and thus energy input, also increases when the
surface area increases. This increase in sample mass will also lead
to more sulfur release, as shown in Figure 6. The sulfur release for
pine wood is less signicant than for tire rubber but also may also
be characterized as linear as a function of surface area.
Effect of Bulk Oxygen Concentration. The sulfur release has
been studied as a function of the inlet oxygen concentration in the
freeboard gas above the rawmaterial bed. The investigation has been
conducted with polypropylene, petcoke, pine wood cubes, and tire
rubber granulate (see Table 1 for details). The results are shown in
Figure 10 with the sulfur release expressed in milligrams of pure
sulfur versus the oxygen concentration expressed in volume percent.
It is observed that the sulfur release increases for all tested fuels when
the oxygen concentration decreases. Particularly, the tire rubber
granulate leads to an increase in sulfur release of nearly a factor 8
when the oxygen concentration changes from 21 to 5 vol % O
2
.
The reason for the high dependency of the oxygen concentra-
tion on the sulfur release becomes clear when the changes in ue
gas composition are considered. Figure 11 shows the combustion
of tire rubber granulate at 21 and 5 vol %, respectively. It is
observed that the CO and SO
2
concentrations reaches appro-
ximately 500 and 350 ppm, respectively, at 21 vol % O
2
, while
the CO and SO
2
concentrations reach approximately 5700 and
2900 ppm, respectively, at 5 vol % O
2
. Thus, the formation of the
reducing agent COis accompanied by a reductive decomposition
of CaSO
4
to SO
2
, and the degree of this reductive decomposition
is directly related to the CO concentration. The same trend was
observed for the other tested fuels.
CONCLUSIONS
Sulfur release from cement raw materials during combustion
of solid fuels has been studied experimentally in a high-tempera-
ture rotary drum experimental setup. The fuels were tire rubber,
pine wood, petcoke, sewage sludge, and polypropylene. The
sulfur release from the raw materials was observed to increase
when (a) the inlet oxygen concentration decreased, (b) the tem-
perature increased, and/or (c) the total surface area of the fuel
particles increased. These three parameters all had the potential
of augmenting local reducing conditions in the raw material bed,
as indicated by elevated levels of CO during the fuel devolatiliza-
tion. The sulfur release increased as a consequence of the re-
ducing conditions, due to the lower stability of CaSO
4
.
The type of fuel also had a signicant eect on the sulfur release.
The sequence was found to be tire rubber granulate > sewage sludge
> pine wood sawdust > petcoke > polypropylene akes. However,
the sulfur release was signicantly lower when the fuel was one tire
rubber particle or one wood particle, indicating that fuel particle size
has a great eect on the sulfur release from the raw materials.
In order to minimize sulfur release from cement raw materials
during solid fuel combustion in the material inlet end of the
rotary kiln, it is recommended to have as high a bulk oxygen
concentration as possible. This could for example be achieved by
a high excess air ratio in the rotary kiln main burner. The temper-
ature of the cement raw materials should be kept as low as
possible during the fuel devolatilization. This can be achieved by
introducing partially calcined raw materials from the calciner
rather than completely calcined raw materials: The very endo-
thermic calcination process can absorp heat from the fuels,
thereby helping to maintain a relatively low bed temperature
during fuel devolatilization in the rotary kiln material inlet.
Finally, use of coarse fuel particles can slow the fuel devolatiliza-
tion rate and thereby the rapid formation of reducing agents.
Thus, use of coarse fuel particles in the material inlet end of cement
rotary kilns is likely to reduce the tendency for local reducing
conditions relative to the situation with ne fuel particles.
AUTHOR INFORMATION
Corresponding Author
*E-mail: arn@kt.dtu.dk. Telephone: +45 45252831.
ACKNOWLEDGMENT
The work described in this paper is part of a research platform
on future cement technology nanced by The Danish National
Figure 10. Corrected accumulated sulfur release as a function of inlet
oxygen concentration for four dierent fuels. Conditions: 5% ll degree,
900 C, 100 NL/min, 6 rpm. Energy input in each experiment was 30 kJ.
Figure 11. Flue gas composition during combustion of 0.81 g (30 kJ) of
tire rubber granulate at 5 and 21 vol % O
2
. Conditions: 5% ll degree,
900 C, 100 NL/min, 6 rpm.
218
3924 dx.doi.org/10.1021/ef200963v |Energy Fuels 2011, 25, 39173924
Energy & Fuels
ARTICLE
Advanced Technology Foundation, Technical University of
Denmark (DTU) and FLSmidth A/S.
REFERENCES
(1) Verein Deutscher Zementwerke (VDZ), Activity Report 2007
2009, Dusseldorf, Germany, http://www.vdz-online.de. Obtained July
1, 2011.
(2) Nielsen, A. R.; Larsen, M. B.; Glarborg, P.; Dam-Johansen, K.
Energy Fuels 2011, 25, 29172926.
(3) Choi, G.-S.; Glasser, F. P. Cem. Concr. Res. 1988, 18, 367374.
(4) Sylla, H.-M. ZKG 1974, 10, 499508.
(5) Hansen, J. P. SO
2
Emissions from Cement Production. Ph.D.
Thesis, Technical University of Denmark, Department of Chemical
Engineering, 2003; ISBN 87-90142-96-9.
(6) Lyngfelt, A.; Leckner, B. Chem. Eng. Sci. 1989, 44 (2), 207213.
(7) Lyngfelt, A.; Leckner, B. Chem. Eng. Sci. 1999, 54, 55735584.
(8) Dam-Johansen, K.; Hansen, P. F. B.; stergaard, K. Chem. Eng.
Sci. 1991, 46 (3), 847853.
(9) Hansen, P. F. B.; Dam-Johansen, K.; Bank, L. H.; stergaard, K.
Proc. Int. Conf. Fluid. Bed Combust. 1991, 1, 7382.
(10) Hansen, P. F. B; Dam-Johansen, K.; stergaard, K. Chem. Eng.
Sci. 1993, 48 (7), 13251341.
(11) Tarelho, L. A. C.; Matos, M. A. A.; Pereira, F. J. M. A. Fuel Process.
Technol. 2005, 86, 13851401.
(12) Barin, I. Thermochemical Data of Pure Substances, 3rd ed.; VCH:
Weinheim, Germany, 1995; ISBN 3-527-28745-0.
219
Devolatilization and Combustion of Tire Rubber and Pine Wood in a
Pilot Scale Rotary Kiln
Anders R. Nielsen,*
,,
Morten B. Larsen,

Peter Glarborg,

and Kim Dam-Johansen

FLSmidth A/S, Vigerslev Alle 77, DK-2500 Valby, Denmark

Department of Chemical Engineering, CHEC Research Centre, Technical University of Denmark (DTU),
DK-2800 Lyngby, Denmark
ABSTRACT: Cement production is highly energy intensive and requires large quantities of fuels. For both economical and
environmental reasons, there is an increasing tendency for utilization of alternative fuels in the cement industry, examples being
tire derived fuels, waste wood, or different types of industrial waste. In this study, devolatilization and combustion of large
particles of tire rubber and pine wood with equivalent diameters of 10 mm to 26 mm are investigated in a pilot scale rotary kiln
able to simulate the process conditions present in the material inlet end of cement rotary kilns. Investigated temperatures varied
from 700 to 1000 C, and oxygen concentrations varied from 5% v/v O
2
to 21% v/v O
2
. The devolatilization time of tire rubber
and pine wood were found to mainly depend on temperature and particle size and were within 40 to 170 s. Rate limiting
parameters for char oxidation of tire rubber and pine wood were found to be bulk oxygen concentration, mass transfer rate of
oxygen, raw material fill degree, raw material characteristics, and temperature. Kiln rotational speed only had a minor effect on
the char oxidation when the raw material bed was in a rolling motion. Initial fuel particle size also influenced the char conversion
time for pine wood char but had no influence on tire char conversion time, because the tire rubber crackled into several smaller
char fragments immediately after devolatilization. The char conversion times were from 40 to 480 s for tire char and from 30 to
1300 s for pine wood char, depending on the conditions. Models for devolatilization and char oxidation of tire rubber and pine
wood are validated against experimental results.

INTRODUCTION
The cement industry is responsible for approximately 2% of the
worlds primary energy consumption.
13
Fuel consumption
accounts for about 3040% of the total cement clinker pro-
duction costs.
4
Due to competition in the cement market,
increasing fossil fuel prices, and environmental concerns,
cement producers have increased the utilization of alternative
fuels as a substitute for fossil fuels in order to achieve the
most economic fuel mix. In this context, alternative fuels
cover all nonfossil fuels and waste from other industries. Popu-
lar alternative fuels in the cement industry are tire-derived
fuels (TDF), biomass residues, and different commercial and
industrial wastes.
It is attractive to utilize coarse, solid alternative fuel particles
directly in the material inlet end of cement rotary kilns in order
to save expenses for shredding of the fuels to smaller particles
and to increase fuel flexibility of the system. High temperatures
in the rotary kiln and material retention times of several
minutes provide good conditions for fuel burnout. Combustion
of solid fuels in the material inlet end may, however, have
negative impact on the process stability of the kiln system in the
form of increased tendency for deposit build-ups. The
mechanism behind formation of deposit build-ups is described
elsewhere.
5
The influence of alternative fuel utilization in
the material inlet end of the rotary kiln on process stability
often limits how large quantities of fuel are possible to use
in this way. Little information is available in the literature
about combustion of large fuel particles under the conditions
in the material inlet end of rotary kilns, and it is possible
that a systematic investigation can optimize the utilization of
alternative fuels in the material inlet end of rotary kilns. Figure 1
shows utilization of large fuel particles in the form of whole
tires in the material inlet end of the rotary kiln. The raw
material bed temperature is typically around 900 C and
increases toward 1450 C in the burning zone. The gas
temperature is around 10001200 C in the material inlet end
and may be above 2000 C in the burner flame in the burning
zone. Bulk oxygen concentration in the material inlet end is
typically quite low, 27% v/v. Solids and gas residence times in
the rotary kiln are typically 1530 min and 510 s, respectively.
Devolatilization of large TDF particles have been studied by
several researchers.
612
In addition, studies have been made
with micro thermogravimetric analysis (micro-TGA) on small
TDF samples to determine reaction kinetics and composition
of oil, gas, and char.
1315
Devolatilization times of cylindrical
tire rubber particles with diameters from 7 to 22 mm and
heights of 35 mm at 840 C in an inert atmosphere are re-
ported to be 75 to 300 s when increasing the particle diameter
from 7.5 to 22 mm.
11
Devolatilization times of tire rubber
particles with thicknesses in the range of 612 mm in air and
in the temperature interval from 700 to 1000 C were found
to be 30100 s depending on thickness and temperature.
12
Generally, results reported in literature about devolatilization of
large TDF particles show clear tendencies for devolatilization
time to increase with increasing particle size and decrease with
increasing temperature and/or oxygen concentration.
Received: September 8, 2011
Revised: December 20, 2011
Published: December 20, 2011
Article
pubs.acs.org/EF
2011 American Chemical Society 854 dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868
220
Previous work about tire char combustion has been made
with a few mg of tire char particles in the 100500 m size in
thermogravimetric analyzers (TGA reactors) and drop tube
reactors.
13,14,1618
Char conversion times are reported to be
well below 1 s in air and at 1200 C for tire char with particle
size of around 500 m.
16
Oxidation time for tire char with
particle size between 102 and 212 m at 10% v/v O
2
were
found to be from 9 s at 750 C to 4 s at 850 C.
18
Masi et al.
found that a TDF char particle with a diameter of 100 m
reached full conversion after 6 s at 5% v/v O
2
and 850 C.
17
According to Masi et al., a medium rank coal char would
require 100 s to reach full conversion under the same
conditions.
17
Masi et al. also compared intrinsic TDF char
reactivity with the intrinsic char reactivity of refuse derived fuels
(RDF) and biomass (Robinia Pseudoacacia) and found that
the TDF char was most reactive.
17
Atal and Levendis also re-
ported that tire char burned 24 times faster than bituminous
coal char.
16
Larsen et al. concluded that the TDF chars were
very reactive at temperatures of 850 C and higher.
18
As a
consequence, intraparticle kinetics were concluded to be less
important because the reaction would take place at the outer
particle surface. In addition, the ash layer formed on the particle
surface was observed to be very porous and could easily be
removed by slight mechanical interaction of the particle with
tweezers. In the rotary kiln, the rotational movement of the
kiln is assumed to remove the ash layer rapidly, leaving an
unconverted char particle.
Devolatilization of large wood particles has also been studied
by several researchers.
1924
Winter et al. measured temperature
profiles in spherical 320 mm diameter beech wood particles at
oxygen concentrations from 0 to 21% v/v and temperatures
from 700 to 950 C. The experiments were made in a fluidized
bed with silica sand as bed material. The devolatilization time of
a 10 mm diameter beech wood particle at 800 C and oxygen
concentration of 10% v/v was approximately 55 s.
19
Devola-
tilization of nonspherical pine wood particles with equivalent
diameters, d
eff
, in the range of 10 to 45 mm and in the
temperature interval of 650 to 850 C has been investigated by
de Diego et al.
21,22
The reported devolatilization times for 10 to
30 mm diameter particles were, for example, from 30 to 150 s at
850 C. The devolatilization was reported to be only slightly
affected by a change in atmosphere from air to nitrogen.
21,22
Di Blasi and Branca studied temperature profiles in cylindrical
beech wood particles with lengths of 20 mm and diameters
from 2 to 10 mm.
23
The experimental setup was a fluidized bed
with sand as bed material and an inert atmosphere of nitrogen.
Bed temperatures were from 439 to 834 C. An empirical
correlation was suggested for the devolatilization time, and the
devolatilization time for a 10 mm diameter beech wood particle
at 834 C was, for example, found to be approximately 40 s.
23
Jand and Foscolo studied the pyrolysis of spherical beech wood
particles with diameters from 5 to 20 mm and in the tem-
perature interval of 560740 C, in a fluidized bed gasifier with
sand as bed material. The devolatilization times at 740 C were
from 40 to 140 s, depending on the particle diameter.
24
The results reported in literature about devolatilization of
large wood particles are as indicated above, typically based on
experiments in fluidized bed reactors with sand as bed material
and at temperatures in the interval of 4001000 C. The wood
types are either beech wood or pine wood. The general
tendencies are that devolatilization times increase with
increasing particle size and decrease with increasing temper-
atures. The effect of the atmosphere seems to be of minor
importance.
Most studies with wood char combustion are made with
wood chars with particle sizes in the m range.
2527
These
studies typically use TGA experiments to obtain information
about char intrinsic kinetics. Jand and Foscolo studied the char
combustion of large spherical beech wood char particles with
diameters from 5 to 20 mm at 740 C and in air in a fluidized
bed reactor.
24
The char oxidation times were found to be
approximately 150 s for a 5 mm diameter char particle and
500 s for a 20 mm diameter char particle.
The aim of this work is to obtain quantitative data on the
devolatilization and combustion of large particles of tire rubber
and pine wood under conditions that resemble those in the
material inlet end of rotary kilns. The purpose is to develop
models for fuel conversion times in cement rotary kilns.

EXPERIMENTAL SECTION
Experimental Setup. The experiments were made in a high
temperature pilot scale rotary kiln in a commercially available chamber
Figure 1. Utilization of large solid alternative fuel particles in the material inlet end of cement rotary kilns. Typical gas and bed temperatures shown
for clarity.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 855
221
furnace; see Figure 2. The setup can reach an operating temperature of
1000 C. It is electrically heated by heating elements embedded in two
side walls and the bottom. The steel kiln is supported by a steel tube
on rollers which passes through the rear wall. This steel tube is
connected to a motor by means of a roller chain, in order to rotate the
kiln. The part of the steel tube that is outside the chamber furnace is
water cooled in order to protect the rollers, roller chain, and gas seals
against the high temperatures.
Gas can be transported to the chamber furnace through a hole in
the roof. The furnace has been designed to achieve sufficient heating of
reactant gas before entering the pilot scale rotary kiln, which was also
verified by temperature measurements. The gas is transported into the
rotating steel kiln due to an externally placed gas pump that pumps the
gas through the rotating kiln and steel tube where after the gas exits
the reactor for subsequent cooling and analysis.
The chamber furnace door is equipped with a window for visual
inspection, two holes for thermocouples, and a centrally placed water
cooled tube for solid fuel addition. The solid fuels are placed in a
sample container which can be pushed into the hot pilot scale rotary
kiln or pulled out to the water cooled tube.
The experimental setup is shown in Figure 3. The gas supply can be
up to 500 NL/min of air and nitrogen. The gas is transported to the
pilot scale rotary kiln reactor during the experiments but may also be
bypassed during calibration of flow controllers, leakage tests, etc.
The temperature is measured in three different positions in the pilot
scale rotary kiln reactor: at the rear wall in the chamber furnace (T1),
in the center of the pilot scale rotary kiln (T2), and at the door just in
front of the pilot scale rotary kiln (T3). A pump transports the flue gas
out of the pilot scale rotary kiln reactor. Before passing the pump,
the flue gas is cooled in a heat exchanger and soot particles are
captured in a filter. After having passed the pump, the flue gas is sent
directly either to the stack or to the gas analyzers for measurement
of O
2
, CO, CO
2
, SO
2
, and NO. The fraction of the flue gas that is
transported to the analyzers will pass another two filters, a gas cooler
for condensation of water, and a sample gas pump before reaching the
gas analyzers.
Materials and Methods. Proximate and ultimate analyses as well
as lower heating values for the tested fuel samples are shown in Table 1.
Tire rubber and pine wood are cut into cylindrical and rectangular
shapes, respectively.
Raw materials are placed in the kiln to obtain the desired volumetric
fill degree. The reactor door is closed, and the reactor is heated to the
desired temperature. The kiln rotates always when the reactor is
heated and keeps the raw materials in a rolling motion, with an angle
Figure 2. High temperature pilot scale rotary kiln. Left: Side view. Right: Top view.
Figure 3. Sketch of the high temperature pilot scale rotary kiln experimental setup.
Table 1. Fuel Analyses and Lower Heating Values (LHV) for
Solid Fuels Used in the Experiments
a
proximate analysis ultimate analysis
VM FC Ash C H N S
LHV
(MJ/kg)
tire
rubber
64.6 32.6 2.8 87.4 7.1 0.3 1.2 36.9
pine
wood
75.3 24.5 0.2 38.9 5.2 0.1 16.2
a
Units are in wt % as received and MJ/kg.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 856
222
of repose of approximately 3040. It is the thermocouple T2 in the
center of the pilot scale rotary kiln that is used to determine the
temperature prior and during the experiments. The fuel sample is
placed in the sample boat, and the sample boat is then positioned in
the water cooled tube. The water cooled tube is closed in the end, in
order to obtain a controlled atmosphere inside the pilot scale rotary
kiln reactor. The gas to the reactor is adjusted to the desired flow
and oxygen concentration. The temperature at T2 and oxygen con-
centration at the O
2
analyzer is monitored. When a stable temperature
and oxygen concentration are reached, the fuel sample boat is pushed
into the pilot scale rotary kiln, turned 180 and pulled out in the water
cooled tube again. The fuel sample will thus drop into the preheated
raw materials and immediately be heated, followed by devolatilization
and char oxidation. Temperatures and flue gas compositions are con-
tinuously logged during the experiments.
An example from a typical experiment with combustion of one tire
rubber cylinder is shown in Figure 4. The oxygen concentration is
observed to decrease from 10 toward 7% v/v when the devolatilization
starts. At the same time, the CO concentration increases from 0 ppmV
to approximately 500 ppmV, and the CO
2
concentration increases
from 0 to 2.5% v/v during the devolatilization. The char oxidation is
assumed to be finished when the CO
2
concentration becomes lower
than 0.04% v/v. This assumption is in good agreement with the visual
observations during the experiment where red-glowing char particles
can be observed in the bed until the CO
2
concentration reaches values
below around 0.04% v/v CO
2
. Furthermore, since the experiments are
made with air/nitrogen mixtures, the CO
2
concentration for experi-
ments with pure air will never decrease below approximately 0.04% v/v,
which is the CO
2
concentration in Earths atmosphere.
28
The degree of fuel conversion against time is derived, assuming that
the fuel conversion is proportional to the carbon conversion, by
integration of the concentration profiles of CO
2
and CO:
18

=
+
+

X t
y y t
y y t
( )
d
d
t
0
CO CO
0
CO CO
2
2
(1)
The repeatability of the experiments is generally found to be
acceptable. Nearly all experiments were repeated at least three times to
ensure the repeatability. The filters in the experimental setup have
been cleaned regularly in order to avoid increases in pressure drop
over the system, which were observed to affect the repeatability of
individual experiments. A sensitivity analysis based on the accumu-
lation law of uncertainties was conducted to estimate the uncertainty
on the degree of fuel conversion.
29
In the estimation of the overall
uncertainty, the temperature is assumed to fluctuate 1%, the gas flow
is assumed to fluctuate 3%, and the uncertainties on each of the
species CO and CO
2
are assumed to fluctuate 1% on each species.
The pressure P is assumed to be constant. The sensitivity analysis
indicates an overall relative uncertainty on the estimated degree of fuel
conversion in the order of 6%.

RESULTS AND DISCUSSION


General Observations. During the experiments, it was possible
to follow the devolatilization and char oxidation visually through
the window. This gave a good impression of the main reaction
pathways for the different fuels. For all the fuels, a flame front
surrounding the fuel particles were observed almost imme-
diately after insertion into the pilot scale rotary kiln, indicating
the start of the devolatilization.
The tire rubber particles were observed to predominantly
keep the particle shape under the devolatilization but with some
degree of fragmentation. After the devolatilization, the tire char
fragmented into several smaller char particles. The pine wood
particles kept their shape during the devolatilization and, to a
large extent, also during the char oxidation.
The experiments have been made with coarse sand as the raw
material. This raw material was chosen because (1) it is inert
and (2) it gives the desired rolling motion of the bed. It was
attempted to use real cement raw material but without success
due to the sticky nature of the raw material which partly
resulted in a thick coating layer on the inner kiln wall and partly
in clinker nodulization.
Effect of Fuel Sample Mass. The effect of fuel sample
mass has been investigated with tire rubber. Figure 5 shows the
conversion time versus mass for tire rubber cylinders with
dimensions D = 9 mm and L = 25 mm and roughly spherical
Figure 4. Example of experiment with combustion of a tire rubber
cylinder. 900 C and 10% v/v O
2
. 5% volumetric kiln fill with coarse
sand, 6 rpm, 100 NL/min. Particle dimensions: D = 9 mm and L =
25 mm.
Figure 5. Conversion time versus mass for tire rubber granulate and
cylinders. 10 % v/v O
2
. Empty pilot scale rotary kiln, 900 C, 6 rpm,
100 NL/min. Conversion times evaluated as 80% carbon conversion.
18
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 857
223
tire rubber granulate particles with D = 2 mm. The masses are
in the interval 1.8 to 15 g, and the results shown are made at
900 C and 10% v/v O
2
and in an empty pilot scale rotary kiln.
The conversion times are evaluated at 80% carbon conversion
rather than at full conversion: This is to avoid the consequences
of dispersion and time delay between the experimental setup
and the gas analyzers which makes the time for full conversion
uncertain.
18
It is observed that the conversion time increases
linearly with mass for the tire cylinders as well as for the tire
granulate. It is furthermore observed that there are no particular
differences in conversion time for the tire cylinders and tire
granulates when the total mass is the same. Thus, the initial
fuel particle size of the tire rubber does not have any particular
effect on the conversion time. This result is in good corre-
spondence with the visual observation that the tire rubber
rapidly cracks into several small fragments after the devolatili-
zation, so conversion time will be a function of overall tire
rubber mass rather than tire rubber particle size. The results are
quite different with pine wood where the conversion time was
observed to be a function of both particle size and sample mass,
indicating that these fuels follow another conversion pathway
than tire rubber.
Effect of Fuel Particle Size and Shape. The effect of fuel
particle size has been studied with cylindrical tire rubber
particles and rectangular wood particles. The rectangular wood
particles are also used to study the effect of fuel particle shape.
Figure 6 left shows the conversion curves for three tire rubber
cylinders of different diameters (and different masses) at
900 C and 10% v/v oxygen. The overall conversion time is
observed to increase when the particle diameter (and mass)
increases. It should be noted that the devolatilization time also
increases when the particle diameter increases. However, on the
basis of the study of fuel sample mass shown in Figure 5, it
must be concluded that the increase in overall tire rubber con-
version time is due to the increase in mass rather than the
increase in particle diameter, since the tire char rapidly frag-
ments into several smaller char particles.
Figure 6 right shows the carbon conversion versus time for
five pine wood particles with different masses, sizes, and shapes.
Four of the particles have a thickness of 10 mm while the fifth
particle has a thickness of 20 mm. It is observed that the
devolatilization time is approximately 1 min for the four 10 mm
thick particles whereas the devolatilization time is around 2 min
for the 20 mm particle; the exact devolatilization times are
shown in the Model Analysis section. The char oxidation times
are generally observed to increase when the mass of char
increases, but it is also observed that the shape influences the
char oxidation time. Three of the particles have approximately
the same mass of 7 g but different char oxidation times: The
longest of the 10 mm thick particles, with dimensions 120
15 10 mm, is fully combusted after approximately 11 min
while the shorter, more compact particles with dimensions 60
30 10 mm and 30 30 20 require approximately 16 and
18 min for full conversion, respectively. Regarding particles
with similar mass but different shapes, the reason for the
difference in conversion times may be explained by the external
surface area of the particles, which is largest for the particle that
reaches full conversion first. Overall, the results indicate that
particle thickness and external surface area have an effect on
both the devolatilization time and the char combustion time.
Effect of Raw Material Bed Fill Degree. The effect of
the raw material bed fill degree on the conversion time has
been investigated. Figure 7 left shows the conversion time for
cylindrical tire rubber particles at volumetric fill degrees of 0%
(empty pilot scale rotary kiln), 5%, and 10% at 900 C and 10%
v/v oxygen. The overall carbon conversion times are observed
to be highly influenced by the fill degree: In an empty pilot
scale rotary kiln, the conversion time is around 100 s while it is
around 200 s at 5% fill degree. For the highest fill degree of
10%, the conversion time is 600 s, significantly longer for the
other two cases. It is also interesting to note that the shape of
the curves are quite different, indicating that the fraction of
volatile carbon and char carbon are different depending on the
kiln fill degree. The reason may be differences in the heating
mechanism, depending on whether the tire cylinder is buried in
the bed or only exposed to the wall and freeboard gas. It has
previously been reported that the heating rate of tire rubber has
Figure 6. Left: Carbon conversion for three different tire rubber cylinders. L
Cylinder
= 25 mm. Right: Carbon conversion for pine wood particles of
different sizes. T = 900 C, 5% fill, coarse sand, 10% v/v O
2
, 6 rpm, 100 NL/min.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 858
224
a significant effect on the devolatilization and char combustion
process.
14
Figure 7 right shows the conversion time for a pine wood
particle at fill degrees of 0% (empty pilot scale rotary kiln), 5%,
and 10%. The devolatilization time is observed to be identical
for the three cases while the char oxidation time increases with
increasing fill degree. Thus, the raw material fill degree is of
importance for the char oxidation time, which is likely to be due
to diffusion limitations of oxygen to the char particle when fully
or partly covered with raw materials.
Effect of Oxygen Concentration. The effect of the
oxygen concentration on the conversion time has been investi-
gated in the interval 5% v/v to 21% v/v. Figure 8 left shows the
conversion time for cylindrical tire rubber particles as a function of
the oxygen concentration. It is seen that the conversion time
increases when the oxygen concentration decreases. The increase is
approximately linear with conversion times of 140, 170, 200, and
380 s at 21, 15, 10, and 5% v/v O
2
.
Figure 8 right shows the conversion time for pine wood
particles as a function of the oxygen concentration. It is
observed that the oxygen concentration has practically no effect
on the devolatilization time but a significant effect on the char
oxidation time. Full conversion is reached in less than 10 min at
the highest oxygen concentrations of 21% and 15% O
2
, re-
spectively. At 10% O
2
, the conversion time is 18 min, and at 5%
O
2
, the conversion time is 40 min. Thus, the oxygen concen-
tration clearly has a significant effect on the conversion time of
pine wood char, and the conversion time increases approx-
imately linearly as the oxygen concentration decreases.
Effect of Rotational Speed. The effect of kiln rotational
speed on the conversion time has been studied. Figure 9 left
shows the conversion time for cylindrical tire rubber particles as
Figure 7. Carbon conversion as a function of raw material fill degree. Left: Tire rubber particle with dimensions: D = 9 mm, L = 25 mm. Right:
Pine wood particle with dimensions: 60 30 10 mm. Coarse sand. T = 900 C, 10% O
2
, 6 rpm, 100 NL/min.
Figure 8. Carbon conversion as a function of oxygen concentration. Left: Tire rubber particle with dimensions: D = 9 mm, L = 25 mm. Right:
Pine wood particle with dimensions: 60 30 10 mm. Coarse sand, 5% fill. T = 900 C, 6 rpm, 100 NL/min.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 859
225
a function of rotational speed in a bed with 5% fill degree at
900 C and 10% v/v O
2
. It is observed that the conversion time
increases when the rotational speed decreases. The conversion
time is, e.g., 100 s at 20 rpm and 200 s at 6 rpm, a factor 2 in
difference. The conversion time increases dramatically to
440 s at the lowest rotational speed of 3 rpm. The reason is
likely to be a change in bed behavior, which could also be
observed visually: The bed motion is slumping at 3 rpm
while it is rolling at 6 rpm and higher rotational speeds. This
difference in bed motion affects the mixing process and thus
the conversion time of the fuel particles. It is also possible
that the fuel particle breakdown is more significant at higher
rotational speeds due to stronger forces acting on the fuel
particles.
Figure 9 right shows the conversion time for pine wood
particles as a function of rotational speed in a bed with 5% fill
degree at 900 C and 10% v/v O
2
. It is observed that there is no
difference for the devolatilization time. For the char oxidation,
the conversion time increases slightly when the rotational speed
is decreased in the interval of 20 to 6 rpm. However, when the
rotational speed is lowered to 3 rpm, the conversion time is
increased significantly (33 min relative to 17 min at 6 rpm).
This large effect at the lowest rotational speed is due to a
change in bed behavior, from rolling to slumping.
Effect of Temperature. The effect of temperature on the
conversion time has been investigated for the temperatures
700 C, 800 C, 900 C, and 1000 C. Figure 10 left shows the
conversion time of tire rubber as a function of temperature. It is
observed that both the devolatilization time and the char oxid-
ation time decreases when the temperature increases. The
conversion time at 1000 C is observed to be around 160 s,
while it is 490 s at 700 C.
Figure 9. Carbon conversion as a function of rotational speed. Left: Tire rubber particle with dimensions: D = 9 mm, L = 25 mm. Right: Pine wood
particle with dimensions: 60 30 10 mm. Coarse sand, 5% fill. T = 900 C, 10% O
2
, 100 NL/min.
Figure 10. Carbon conversion as a function of temperature. Left: Tire particle with dimensions: D = 9 mm, L = 25 mm. Right: Pine wood particle
with dimensions: 60 30 10 mm. Coarse sand, 5% fill. 6 rpm, 10% O
2
, 100 NL/min.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 860
226
Figure 10 right shows the conversion time for pine wood
particles as a function of temperature in the temperature
interval of 700 C to 1000 C. Both the devolatilization
time and the char oxidation time increases when the tem-
perature decreases. However, the time difference for
the char oxidation at 700 C and 800 C is quite small, 0
to 1 min, whereas the time is significantly shorter at 900 C
and 1000 C.
Conclusions on Experimental Parameter Study. A
parameter study has been conducted for tire rubber and pine
wood. The main findings were:
For pine wood, fuel mass, particle thickness, and external
surface area are important for the conversion time, both
regarding devolatilization and char oxidation. For tire rubber,
however, the tire char rapidly fragments into smaller particles.
Consequently, the initial dimensions of the tire rubber are
of little importance for the tire char conversion time. For
the char oxidation part of tire rubber and pine wood, the
parameters char mass, oxygen concentration, temperature,
and raw material fill degree are all important parameters, but
also rotational speed was observed to have an effect on the
conversion time.

MODEL ANALYSIS
This section seeks to develop and verify mathematical models
for devolatilization and char oxidation of selected alternative
fuels. The models are to be used under conditions similar to
those in the material inlet end of cement rotary kilns.
Model for Devolatilization. The heat up of a large,
spherical particle may be derived by solving the unsteady heat
transfer differential equation:
30,31

T
t
r
r
r
k
C
T
r
1 p
2
2
p
p
p
p
(2)
with the following boundary and initial conditions:

= +
+
+ =
k
T
r
Yh T T Y
h T T T T
T T r R
( ) (1 )
( ( ) ( )
( )) for
p
p
Cond b p
Conv g p p g g
4
g p
4
p w w
4
w p
4
(3)

= =
T
r
r 0 for 0
p
(4)
where R is the initial radius of the particle, k
p
is the thermal
conductivity of the particle,
p
is the virgin fuel density, and
C
p
is the specific solid heat capacity. h
Cond
and h
Conv
are the heat
transfer coefficients for conduction and convection, respec-
tively. T
p
, T
b
, T
g
, and T
w
are temperatures of particle, bed, bulk
gas, and inner kiln walls, respectively. Y is the dimensionless
distribution of fuels in the bed, which is 1 if the fuel particles
are fully covered by raw materials and 0 if the fuel particles are
fully exposed to the gas phase. Equation 3 expresses that all
heat transferred to the surface of the particle is conducted into
the particle, and eq 4 expresses that the gradient at the particle
center is zero due to symmetry. Equation 3 must be solved
numerically. However, if radiation is neglected, an analytical
solution can be found.
31

T T
T T
2
sin cos
sin cos
sin
e
i i
r
r
r
r
a p
a p,0
1
i i i
i i
i
i
( t/r )
p
p,0
p
p,0
i
2
p,0
2
(5)
where T
a
is the mean average temperature surrounding the
particle. T
a
may be estimated as T
a
= Y T
b
+ (1 Y) T
g
. is
the thermal diffusivity, = k
p
/(
p
C
p
).
The
i
s are the positive solutions to the equation:
= Bi cos (1 ) sin (6)
where Bi is the Biot number. An efficient heat transfer
coefficient, h
eff
, is used to calculate Bi, where h
eff
= Y h
Cond
+
(1 Y) (h
Conv
+ h
rad
) and h
rad
=
g
(T
g
2
+ T
p
2
) (T
g
+ T
p
).
32
The time,
devol
, required for complete devolatilization
corresponds to the time needed for the fuel particle center to
reach the upper devolatilization temperature, T
Vol2
. This time is
determined from eq 5 using T
p
(0, t) = T
Vol2
.
33
However, eq 5
should only be used to determine the devolatilization time for
an initially dry particle, since the equation does not take the
particle moisture content into consideration. The moisture
content of a fuel particle may influence the devolatilization time
considerably.
21,22,31,33
Validation of Devolatilization Model. If the temperature
required for complete devolatilization, T
Vol2
, is known, e.g.,
from TGA measurements reported in the literature, eq 5 may
be used to determine the time required for full devolatilization.
Table 2 shows the temperatures for initial and final
devolatilization temperatures for the fuels tire rubber and
pine wood.
Experiments with nonspherical tire rubber and pine wood
particles of different size and shapes have been conducted to
study the devolatilization time. Since the model assumes that
the fuel particles are spherical, it is required to represent the
nonspherical fuel particles as spherical particles of diameter, d
eff
,
characterized by the nonspherical and spherical particles having
the same total surface area:
35
= d d
eff sph (7)
where d
sph
is the equivalent spherical diameter, defined as the
diameter of a sphere with the same volume as the particle. is
the sphericity defined as:
=

Surface of sphere
Surface of particle
of same volume
(8)
The experimentally found devolatilization times are
determined by means of the normalized carbon conversion
curves for the fuels. Figure 11 shows an example with pine
wood, where the devolatilization time is observed to begin at
Table 2. Temperatures for Initial and Final Devolatilization
of Tire Rubber and Pine Wood
fuel T
Vol1
[K] T
Vol2
[K] reference
tire rubber 523 773 12
pine wood 473 773 34
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 861
227
time 0 s and end at time 80 s. The devolatilization ends when
the slope of the conversion curve is observed to decrease
significantly, indicating the start of the slower char oxidation.
Devolatilization of Tire Rubber. A comparison between
experimental values and values predicted by the model for tire
rubber is shown in Figure 12. Values used in the model are
summarized in Table 3.
In the model, the fraction of the fuel particle surface that is
covered by raw materials, Y, is 0.2. This fraction was chosen on
the basis of visual observations during the experiments, where
the cylindrical tire rubber particle was observed to be free of the
raw material bed during most of the devolatilization. The
experimental values were obtained for four different tire rubber
particle diameters from 10 mm to 18.5 mm and at 900 C. The
two smallest tire rubber particles were also studied at 700 C,
800 C, and 1000 C. The model is in acceptable agreement
with the experimental values, which were found to have
devolatilization times from approximately 40 to 120 s,
depending on temperature and tire rubber particle size. The
model generally overestimates the effect of temperature
compared to the experimental findings: For the two smallest
fuel particles, the difference in experimentally found devolati-
lization times were smaller than that predicted by the model.
Chinyama and Lockwood studied the devolatilization times
of tire rubber particles with thicknesses in the range of 612
mm and in the temperature interval from 700 to 1000 C.
12
The devolatilization times were found to be 30100 s,
depending on thickness and temperature, which were typically
20 s longer than the values found in this study for the same
particle sizes and temperatures. This difference in time may
be due to different heat transfer mechanisms at the two ex-
perimental conditions: In the experiments conducted by
Chinyama and Lockwood, heat transfer to the tire rubber
particles were exclusively by convection and radiation, whereas
heat transfer in this study was also partly by conduction from
the raw materials and kiln wall. Another reason may be the
mechanical influence from the pilot scale rotary kiln in this
study, which is likely to increase fragmentation of the fuel
particle during devolatilization; in fact, the tire rubber particles
were observed to fragment a little during the experiments,
thereby leading to smaller fuel particle sizes.
Larsen et al. also studied the devolatilization of cylindrical
tire rubber particles with diameters from 7 to 22 mm and
heights of 35 mm at temperatures up to 840 C in an inert
atmosphere.
11
The reported devolatilization times at 840 C
were 75 to 300 s when increasing the particle diameter from
7.5 to 22 mm. These devolatilization times are approximately
twice as long as indicated by the results from this study. The
reason for the deviations may again be differences in the heat
transfer mechanisms due to different experimental conditions
as well as the mechanical influence in this study. In addition,
the experiments conducted by Larsen et al. were made under an
inert atmosphere while the present study was made under
oxidizing conditions which are likely to have accelerated the
devolatilization.
Devolatilization of Pine Wood. A comparison between
experimentally found values and values predicted by the model
for pine wood is shown in Figure 13. Values used in the model
can be found in Table 3. The experimental values were
obtained for five different pine wood particle diameters from
12 mm to 26 mm and at 900 C. The pine wood particles of
diameters 15 mm and 20 mm were also studied at 700 C, 800 C,
Figure 11. Determination of devolatilization time for a pine wood
particle at 900 C and 10% v/v O
2
. 5% volumetric fill, coarse sand,
100 NL/min, 6 rpm. Particle dimensions: 60 30 10 mm (7.0 g).
Figure 12. Comparison of tire rubber devolatilization time predicted
by the model and by experimentally found values. 10% O
2
. 5% volu-
metric fill degree in pilot scale rotary kiln. Y = 0.2. = 1.25 10
7
m
2
/s.
Table 3. Fuel Specific Data Used in the Devolatilization
Models
property TDF
pine
wood references
specific heat capacity, C
p
, [J/(kg K)] 2000 1500 11
36
initial particle density,
p
, [kg/m
3
] 1150 690 8
36
k
p
, virgin fuel thermal conductivity,
[W/(m K)]
0.3 0.2 37
36
conductive heat transfer coefficient, h
CBP
,
[W/(m
2
K)]
250 250 38
convective heat transfer coefficient, h
CBP
,
[W/(m
2
K)]
150 150 39
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 862
228
and 1000 C. The model is observed to be in acceptable
agreement with the experimental values, which were found to
have devolatilization times from approximately 40170 s, de-
pending on temperature and particle size. However, the
experiments conducted at 700 C and 800 C are generally
2030 s longer than the devolatilization times predicted by the
model, and in one case, with a particle diameter of 20 mm and
at 700 C, the deviation is 50 s.
Devolatilization of nonspherical pine wood particles with
equivalent diameters, d
eff
, in the range of 10 mm to 45 mm and
in the temperature interval of 650 C to 850 C has been
investigated by de Diego et al.
21,22
The reported devolatilization
times were in good correspondence with the values found in
the present study. For example, at 850 C, the devolatilization
times for 10 mm to 30 mm diameter particles were reported to
be from 30 s to 150 s, whereas the results from this study
indicates devolatilization times from 25 s to 180 s at 850 C.
These minor differences in devolatilization times may be due to
different characteristics of the pine wood used in the
experiments. The found devolatilization times for pine wood
are also in relatively good consistency with devolatilization
times for beech wood particles of similar size as reported by
Di Blasi and Branca
23
and Jand and Foscolo;
24
see the literature
study in the Introduction.
Model for Char Combustion. The aim of this section is to
derive a model for char conversion in the pilot scale rotary kiln,
including the main rate limiting parameters for char conversion.
The char combustion model will be validated against experi-
mental results. The model may be modified into a realistic char
conversion model for the material inlet end of industrial rotary
kilns. The char conversion model may be used in combination
with models for drying and devolatilization of virgin fuel
particles, or it may stand alone to simulate the fate of char
particles dropping from calciners through the kiln riser duct and
into the material inlet end of rotary kilns.
Figure 14 shows a rotary kiln where char particles are buried
in the raw material bed. The bulk oxygen volume fraction is
denoted y
O2,
, and the oxygen volume fraction at the bed
surface is called y
O2,surf
. The oxygen volume fraction at the char
particle surface is called y
O2,part
. The mixing of large particles
into a bed of smaller particles has previously been shown to be
a fast process, typically taking only a few bed revolutions.
40
In a cement rotary kiln, it may be assumed that mass transfer
of oxygen into the active layer is fast while oxygen mass transfer
into the passive layer takes place by a relatively slow diffusion;
see Figure 14. Since the char particles will continuously be
transported through the active and passive layers, respectively,
it may be assumed that char oxidation primarily takes place
when the char particles are present in the active layer.
With the above-mentioned considerations in mind, a model
for char oxidation applicable for the conditions in the material
inlet end of a rotary kiln should include the following: (1)
external mass transfer of oxygen to bed surface, (2) mass
transfer of oxygen through the active layer, (3) diffusion of
oxygen into the char particle, and (4) chemical reaction of char
and oxygen.
The external mass transfer of oxygen to the bed surface can
be written as the mass transfer coefficient k
g
multiplied by the
driving force:
41
=

r k
P
RT
y y ( )
mol O
m s
g
g
O , O ,surf
2
2
2 2
(9)
where k
g
is found from a Sherwood correlation. For the pilot
scale rotary kiln, where the gas flow is in the laminar regime, the
following correlation is used
42
= = Sh
k L
D
Re Sc 0.646
g
O
1/2 1/3
2
(10)
where Re is calculated from the pilot scale rotary kiln length, L.
This correlation is valid for laminar flow over a flat plate, which
is roughly similar to the situation with gas flow above the bed
surface.
In case of a pilot scale rotary kiln without raw materials and
few fuel particles, the mass transfer coefficient may be found
from a Sherwood correlation for flow past single spheres:
42
=

= +

Sh
k d
D
Re Sc
Re Sc
2 0.6
2 800, 0.6 2.5
g p
O
d
1/2 1/3
d
2
p
p (11)
The mass transfer of oxygen from the bed surface into the
active layer can be written as the mass transfer coefficient in the
active layer, k
AL
, multiplied by the driving force and the
probability that the char particles are in the active layer, P
AL
, at
Figure 13. Comparison of pine wood devolatilization time predicted
by the model and by experimentally found values. 10% O
2
, 5% volu-
metric fill degree in kiln. Y = 0.2. = 1.9 10
7
m
2
/s.
Figure 14. Oxygen mass transfer to char particles in a bed of cement
raw materials in a rotary kiln.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 863
229
any given time:
=

r P k
P
RT
y y ( )
mol O
m s
AL AL
b
O ,surf O ,part surf
2
2
2 2
(12)
The probability for the char particles to be in the active layer
may be estimated by the experimentally determined correlation
suggestion by eq 13, which provides a correlation for the
percentage of visible fuel particles at the bed surface:
40
=

P
V
V D LF
Fr
1
100%
AL
1.5
fuel
fuel
4
2 1.2
0.09
rawmaterial
fuel
0.77
(13)
This correlation may be used since the active layer thickness
is quite small, and it therefore may be assumed that the number
of fuel particles in the active layer is practically the same as the
number of fuel particles that are visible above the bed surface.
A bulk density of 1200 kg/m
3
is used for the raw material.
The dimensionless Froude number is defined as:
=

( )
Fr
R
9.81m/s
2
60
2
2
(14)
The mass transfer coefficient in the active layer, k
AL
, has been
determined by Heydenrych:
43
=

= k D
Sh D
d
Sh
6(1 )
, 3.8
AL O ,eff
O
p
2 2
2
(15)
where D
O2, eff
is the effective diffusion coefficient of oxygen:
=

D D
O ,eff O
2 2
(16)
D
O
2
is the diffusion coefficient for oxygen in air, is the bed
tortuosity, and is the bed porosity or void fraction available
for gas flow. Dias et al. found that the tortuosity, , of a mixed
bed of granular particles could be described as = 1/
n
, where
n depends on the particle packing but usually is in the range of
0.40.5.
44
The rate of oxygen diffusion into the char particle may be
estimated by the expression:
=

r
D
r t
P
RT
y y
( )
( )
mol O
m s
O ,part
p p
O ,part surf O ,part core
2
2
2
2 2
(17)
where D
O2, part
is the intraparticle oxygen diffusion coefficient,
r
p
is the particle radius, y
O2, part surf
and y
O2, part core
are the
oxygen volume fractions at the particle surface and at the
intraparticle ash/char interface, respectively. D
O2, part
may be
calculated from:
=

D D
y y
D
1
O ,part
part
part
O ,eff
Vol ash
part
O ,eff
2 2
2
(18)
where the char particle porosity,
part
, is calculated by
subtraction of the weight fractions of volatiles and ash,
y
Vol
and y
ash
. The char particle tortuosity for chars is typically
around 3 to 5.
45
The rate of reaction at the unreacted char outer surface
between char and oxygen is found from the intrinsic rate
expression multiplied with the total moles and divided by
the char surface area and product distribution ratio between
CO and CO
2
:
=

r
n
A
k
E
RT
y
X
exp
(1 )
mol O
m s
C
char
0
a
p
O ,part core
n
m 2
2
2
(19)
where is the product distribution ratio between CO and CO
2
,
X is the char fractional conversion degree, and A
char
is the char
surface area. A correlation between temperature and the CO/
CO
2
ratio, , during combustion of different coal chars has been
reported by Arthur:
46
= =

R T
CO
CO
10 exp
51916J/mol
2
3.4
(20)
This correlation may be used to estimate at different tem-
peratures under the assumption that it is representative for
other chars than coal chars.
At pseudo steady state, the O
2
flux balances the consumption
at the char particle surface and the char conversion may be
expressed as:
=

n
t
A r
s
d
d
mol C
C
surf
(21)
The conversion may also be expressed in terms of fractional
conversion, X, instead of moles, n
C
, by inserting:
= X
n
n
1
C
C,0 (22)
and
= X
n
n d
1
d
C,0
C
(23)
The following expression is then obtained:
=
X
t
A r
n
d
d
surf
C,0 (24)
The unknown oxygen concentrations y
O2,surf
, y
O2,part surf
, and
y
O2,part core
may be found by solving the four eqs 9, 12, 17, and
19 for the four unknowns r, y
O2,surf
, y
O2,part surf
, and y
O2,partcore
.
This is easily done for n = 1 in eq 19, whereas a numerical
solution procedure is required for n 1. For n = 1, the
following expression for char conversion is derived
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 864
230
=

+ + +

X
t
A
n
y
d
d R T
k P
R T
P k P
r X R T
D P
A
n X k
surf
C,0
O ,
(1 )
(1 ) exp
E
R T
2
g
g
b
AL AL
p,0
1/3
p
O
2
,part
char
C,0
m 1
0
a
p (25)
This expression includes the resistances external oxygen
transport, oxygen transport into the bed, diffusion into the char
particle, and chemical reaction between char and oxygen. The
intraparticle diffusion term may be omitted for small char
particles or for char particles that rapidly fragment into many
smaller char particles. It may also be shown that the term for
char oxidation kinetics becomes negligible at high temperatures
above 900 C. Fuel specific, kinetic data, and densities used in
the model are summarized in Table 4.
Analysis and Comparison of Char Combustion Model.
This section analyses the char combustion model under the
experimental conditions in the pilot scale rotary kiln and
compares the model predictions with experimental results. The
purpose is to study effects of mass, oxygen concentration, bed
fill degree, and temperature on the char conversion time.
Figure 15 left shows the char conversion times for three
different pine wood char particles of masses of 0.12 g, 0.24 g,
and 0.88 g, respectively. The initial fuel particle dimensions are
30 10 8 mm, 30 10 15 mm, and 60 30 10 mm,
respectively. The oxygen concentration is 5% v/v, 10% v/v, or
21% v/v O
2
. It is observed that the model, shown with solid
lines, predicts a linear relationship between conversion time,
char mass, and oxygen concentration. The experimental data
also confirm this linear relationship: The conversion time for
the 0.88 g char particle is, for example, around 300 s, 600 s, and
1300 s at oxygen concentrations of 21% v/v, 10% v/v, and
5% v/v O
2
. The model is generally observed to give a good fit
to the experimental data in the experiments with different char
masses and oxygen concentrations.
Figure 15 right shows the char conversion time as a function
of volumetric bed fill degree and temperature. The fill degrees
are 0% (only char in the reactor), 5%, or 10% and the tem-
peratures are 700 C, 800 C, 900 C, or 1000 C. It is
observed that the char conversion time increases when the bed
fill degree increases. The model shows the same tendencies in
temperature dependence as the experimental results. However,
the experimental data show greater differences in conversion
time as a function of temperature compared to the model,
particularly at 700 C where the deviation is up to 350 s at 10%
fill degree corresponding to a deviation of around 30%.
Figure 16 left shows the char conversion time for three
different tire rubber char cylinders with masses of approx-
imately 0.6 g, 1.4 g, and 2.0 g, respectively. The oxygen
concentration is 5% v/v, 10% v/v, or 21% v/v O
2
. The mo-
del, shown with solid lines, predicts a linear increase in con-
version time with char mass and a linear decrease with oxygen
Table 4. Fuel Specific Data Used in the Char Oxidation
Models
fuel
property TDF pine wood references

p
[kg/m
3
] 1100 690 11
36
k
0
[1/s] 1.5 (5.1) 10
10
1.19 10
8
18
47
E
a
[kJ/mol] 193 28 140 10 18
47
exponent m 0.63 0.89 18
47
LHV [MJ/
kg]
37 14 from
analysis
Figure 15. Comparison of experimental data with model predictions for pine wood. Left: Effect of char mass and oxygen concentration on the char
conversion time. 900 C, 5% fill degree, 100 NL/min. Initial particle dimensions: 30 10 8 mm, 30 10 15 mm and 60 30 10 mm. Right:
Effect of raw material fill degree and temperature on model predicted char conversion time for a pine wood char particle with initial dimensions of
60 30 10 mm. Char mass = 0.88 g. 10 % v/v O
2
, 100 NL/min. Char conversion evaluated at 80% conversion degree.
18
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 865
231
concentration. For example, the predicted conversion time for
2 g of char is observed to be halved from 570 to 280 s when the
oxygen concentration is doubled from 5% v/v to 10% v/v.
Thus, the model predicts that external mass transfer of oxygen
from gas to fuel particle is a main rate limiting parameter. The
model gives the best agreement with the experimental values at
10% v/v and 21% v/v O
2
, where the deviation is generally
within 50 s, but the model fails to give good agreement
with the experiments at 5% v/v O
2
and char masses of 1.4 g and
2.0 g. In these cases, the deviations are up to nearly 30%.
Figure 16 right shows the predicted effect of raw material fill
degree in the interval of 010% and in the temperature interval
of 7001000 C for a cylindrical tire rubber char particle.
Experimental values are also compared with the model. Both
experimental values and the model show that the char con-
version time increases with volumetric bed fill degree and
decreases when the temperature increases. The model only fits
the experimental values in the case with a fill degree of 0%,
corresponding to a situation with only char in the pilot scale
rotary kiln and no cement raw materials. However, the exper-
imental values at 700 C and 0% fill show longer conversion
times than predicted by the model: 120130 s relative to 65 s,
respectively. At 5% fill degree, the model only fits the ex-
perimental values at 900 C and 1000 C but underestimates
the conversion times at 700 C and 800 C by a factor of 2, and
at a fill degree of 10%, the model underestimates the effect of
fill degree and temperature by a factor of 2.5 to 3, depending on
temperature. The effect of fill degree seems to increase more
with fill degree for tire rubber than for pine wood. The reason
for these differences may be related to the chemical and
physical properties of the different chars: Tire char fragments
into several small particles while the wood char to a high degree
retains its size and shape.
Conclusions on Char Conversion Model. A model for
char conversion has been suggested and compared with ex-
perimental data for large particles of tire char and pine wood
char. The model includes the four resistances: external oxygen
diffusion to raw material bed, oxygen transport into the bed,
intraparticle oxygen diffusion, and chemical reaction at the char
surface.
The model is in good correspondence with the observed
conversion times for pine wood char. However, it under-
estimates the effect of bed fill degree for tire char. The model
also underestimates the effect of the lowest temperatures of
700 C and 800 C on the conversion time for both pine wood
char and tire char. Regarding the effect of material fill degree on
the char conversion time, it is required to make more experi-
ments with different fuel chars and particle sizes/shapes to
obtain a better basis for model validation.

CONCLUSIONS
Experiments with tire rubber and pine wood in a high temper-
ature pilot scale rotary kiln have been conducted in order to
quantify the effect of key process parameters such as bed fill
degree, kiln rotational speed, temperature, etc. on the fuel
conversion time. The process conditions have been chosen to
simulate the process conditions in the material inlet end of a
modern cement rotary kiln. Investigated temperatures varied
from 700 C to 1000 C, oxygen concentrations varied
from 5% v/v O
2
to 21% v/v O
2
, volumetric raw material bed fill
degrees varied from 0% to 10%, and kiln rotational speeds
varied from 3 rpm to 20 rpm.
The results showed that devolatilization of tire rubber and
pine wood was mainly influenced by the temperature and fuel
particle size. The devolatilization times were from 40 s for a 10 mm
diameter tire rubber particle at 1000 C to 170 s for a 20 mm
diameter pine wood particle at 700 C. The char oxidation was
influenced by all investigated parameters fuel sample mass,
particle size, temperature, oxygen concentration, bed fill degree,
and rotational speed. Rotational speed, however, was only of
importance if there was a shift in bed motion, e.g., from
slumping to rolling. For changes in rotational speed where the
Figure 16. Left: Comparison between model predictions and experimental data for char conversion time of three different tire rubber char cylinders
with lengths of 25 mm and initial diameters of 9 mm, 12 mm, and 16 mm, respectively. 900 C, 5% raw material fill, 100 NL/min. Char masses and
oxygen concentrations indicated on figure. Right: Effect of raw material fill degree and temperature on model predicted char conversion time for tire
char cylinder with a length of 25 mm and initial diameter of 9 mm. Char mass = 0.6 g. 10 % v/v O
2
, 100 NL/min. Char conversion evaluated at 80%
conversion degree.
18
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 866
232
dominant bed motion was maintained as the rolling motion, the
effect of rotational speed was of minor importance.
The tire char was observed to fragment into smaller particles
during the char oxidation, leading to shorter conversion times
than for pine wood char which predominantly behaved as one
shrinking particle during the char oxidation. The found char
conversion times were from 40 s to 480 s for tire char and from
30 s to 1300 s for pine wood char, depending on the conditions.
Devolatilization and char oxidation models have been deve-
loped and compared with experimental results and published
data. These models may be further modified to predict fuel
conversion times in industrial cement rotary kilns.

AUTHOR INFORMATION
Corresponding Author
*E-mail: arni@flsmidth.com. Telephone: +45 36181970.

ACKNOWLEDGMENTS
The work described in this Article is part of a research platform
on future cement technology financed by The Danish National
Advanced Technology Foundation, Technical University of
Denmark (DTU), and FLSmidth A/S.

NOMENCLATURE
A = Surface area (m
2
)
Bi = Biot number (-)
C
p
= Specific heat capacity (J/(kgK))
D = Diameter (mm or m)
D
O2
= Diffusion coefficient (m
2
/s)
E
a
= Activation energy (J/mol)
F = Fill degree (-)
FC = Fixed Carbon (wt %)
Fr = Froude number (-)
h = Heat transfer coefficient (W/(m
2
K))
k = Mass transfer coefficient (m/s)
k
p
= Thermal conductivity (W/(mK))
k
0
= Pre-exponential factor (1/s)
L = Length (m)
LHV = Lower Heating Value (MJ/kg)
n = Moles (Mole)
P = Pressure (Pa) or probability (%)
r = Radius (mm or m) or rate (mol/m
2
s)
R = Universal gas constant (J/(molK)
Sc = Schmidt number (-)
Sh = Sherwood number (-)
t = Time (s or min)
T = Temperature (C or K)
Y = Distribution between bed and gas (-)
X = Degree of conversion (-)
V = Volume (m
3
)
VM = Volatile Matter (wt %)
= Thermal diffusivity (m
2
/s)
= Positive solutions (-)
= Emissivity or Porosity (-)
= StefanBoltzmanns constant (5.7210
8
W/(m
2
K
4
))
= Density (kg/m
3
)
= Tortuosity (-)
= CO/CO
2
distribution ratio (-)
= Rotational speed (min
1
)
Subscripts
a = Average
AL = Active Layer
b = Bed
cond = Conduction
conv = Convection
eff = Effective
g = Gas
p = Particle
part = Particle surface
rad = Radiation
sph = Spherical
surf = Surface
Vol = Volatile
w = Wall
= Surroundings

REFERENCES
(1) VDZ, Activity Report 20072009; 2009, http://www.vdz-online.
de. Obtained April 20, 2011.
(2) Cembureau, Activity Report 2008; http://www.cembureau.be.
Obtained April 20, 2011.
(3) BP Statistical Review of World Energy; June 2009, http://www.bp.
com. Obtained April 20, 2011.
(4) Cembureau, Alternative Fuels in Cement Manufacture; 1997,
http://www.cembureau.be. Obtained April 20, 2011.
(5) Nielsen, A. R.; Larsen, M. B.; Glarborg, P.; Dam-Johansen, K.
Energy Fuels 2011, 25, 29172926.
(6) Bouvier, J. M.; Charbel, F.; Gelus, M. Resour. Conserv. 1987, 15,
205214.
(7) Yang, J.; Tanguy, P. A.; Roy, C. AIChE J. 1995, 41 (6), 15001512.
(8) Yang, J.; Tanguy, P. A.; Roy, C. Chem. Eng. Sci. 1995, 50 (12),
19091922.
(9) Schmidthals, H. Luftvergasung von Altreifen zur integrierten
stofflichen und energetischen Nutzung im Klinkerbrennprozess; Doktor-
Ingenieur Dissertation, Fakulta t fu r Maschinenbau, Ruhr-Universita t
Bochum, Germany, 2001.
(10) Giddings, D.; Pickering, S. J.; Simmons, K.; Eastwick, C. N.
J. Energy Inst. 2002, 75, 9199.
(11) Larsen, M. B.; Schultz, L.; Glarborg, P.; Skaarup-Jensen, L.;
Dam-Johansen, K.; Frandsen, F.; Henriksen, U. Fuel 2006, 85, 1335
1345.
(12) Chinyama, M. P. M; Lockwood, F. C. J. Energy Inst. 2007, 80 (3),
162167.
(13) Conesa, J. A.; Font, R.; Fullana, A.; Caballero, J. A. Fuel 1998, 77
(13), 14691475.
(14) Leung, D. Y. C; Wang, C. L. J. Anal. Appl. Pyrolysis 1998, 45,
153169.
(15) Kyari, M.; Cuncliffe, A.; Williams, P. T. Energy Fuels 2005, 19,
11651173.
(16) Atal, A.; Levendis, Y. A. Fuel 1995, 74 (11), 15701581.
(17) Masi, S.; Salatino, S.; Senneca, O. Fluidised Bed Combust. 1997,
1, 135143.
(18) Larsen, M. B.; Hansen, M. L.; Glarborg, P.; Skaarup-Jensen, L.;
Dam-Johansen, K.; Frandsen, F. Fuel 2007, 86, 23432350.
(19) Winter, F.; Prah, M. E.; Hofbauer, H. Combust. Flame 1997, 108,
302314.
(20) Di Blasi, C. Chem. Eng. Sci. 2000, 55, 59996013.
(21) De Diego, L. F.; Garca-Labiano, F.; Gaya n, P; Abad, A.; Ada nez,
J. J. Anal. Appl. Pyrolysis 2002, 65, 173184.
(22) De Diego, L. F.; Garca-Labiano, F.; Gaya n, P; Abad, A; Ada nez,
J. Ind. Eng. Chem. Res. 2002, 41, 36423650.
(23) Di Blasi, C.; Branca, C. Energy Fuels 2003, 17, 247254.
(24) Jand, N.; Foscolo, P. U. Ind. Eng. Chem. Res. 2005, 44, 5079
5089.
(25) Janse, A. M. C.; de Jonge, H. G.; Prins, W.; van Swaaij, W. P. M
Ind. Eng. Chem. Res. 1998, 37, 39093918.
(26) Senneca, O. Fuel Proc. Technol. 2007, 88, 8797.
(27) Shen, D. K.; Gu., S.; Luo, K. H.; Bridgwater, A. V.; Fang, M. X.
Fuel 2009, 88, 10241030.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 867
233
(28) Tans, P.; Keeling, R. Trends in Atmospheric Carbon Dioxide;
NOAA/ESRL, www.esrl.noaa.gov/gmd/ccgg/trends/. Obtained December
9, 2011.
(29) Larsen, M. O.; Hellesen, B.; Statistik 1, 3
rd
ed.; Institute of
Applied Chemistry, Technical University of Denmark, Lyngby,
Denmark, 1998 (in Danish).
(30) Jensen, A. Heating and devolatilization of coal particles; Course
note, Department of Chemical Engineering, Technical University of
Denmark, Lyngby, Denmark, 1997.
(31) Agarwal, P. K.; Genetti, W. E.; Lee, Y. Y. Chem. Eng. Sci. 1986,
41 (9), 23732383.
(32) Szekely, J.; Ewans, J. W.; Sohn, H. Y. Gas-solid reactions;
New York: Academic Press, 1976; Vol. 51.
(33) Agarwal, P. K.; Genetti, W. E.; Lee, Y. Y. Fuel 1984, 63, 1748
1752.
(34) Kim, S.-S.; Kim, J.; Park, Y.-H; Park, Y.-K. Bioresour. Technol.
2010, 101, 97979802.
(35) Kunii, D.; Levenspiel, O. Fluidization Engineering, 2
nd
ed.;
Butterworths: Boston, MA, USA, 1991; ISBN: 0-409-90233-0.
(36) Leon, G.; Cruz-de-Leon, J.; Villasenor, L. Holz als Roh- und
Werkstoff 2000, 58, 241246.
(37) Sellassie, K. G.; Moo-Young, H. K.; Lioyd, T. B. Int. J. Environ.
Waste Manage. 2007, 1 (2/3), 179191.
(38) Linjewile, T. M.; Hull, A. S.; Agarwal, P. K. Chem. Eng. Sci. 1993,
48 (21), 36713675.
(39) Tscheng, S. H.; Watkinson, A. P. Can. J. Chem. Eng. 1979, 57,
433443.
(40) Nielsen, A. R.; Aniol, R. W.; Larsen, M. B.; Glarborg, P.;
Dam-Johansen, K. Powder Technol. 2011, 210, 273280.
(41) Larsen, M. B. Alternative fuels in Cement Production. Ph.D.
Thesis, Department of Chemical Engineering, Technical University of
Denmark, Lyngby, Denmark, 2007; ISBN: 978-87-91435-49-8.
(42) Green, D. W.; Perry, R. H. Perrys Chemical Engineers Handbook,
8th ed., McGraw-Hill: China, 2008; ISBN: 978-0-07-142294-9.
(43) Heydenrych, M. D. Modelling of rotary kilns. PhD Thesis,
University of Twente, The Netherlands, 2001; ISBN: 90-36515440.
(44) Dias, R.; Teixeira, J. A.; Mota, M.; Yelshin, A. Sep. Purif. Technol.
2006, 51, 180184.
(45) Johnsson, J. E.; Jensen, A. Proc. Combust. Inst. 2000, 28, 2353
2359.
(46) Arthur, J. R. Trans. Faraday Soc. 1951, 47 (2), 164178.
(47) Kastanaki, E.; Vamvuka, D. Fuel 2006, 85, 11861193.
Energy & Fuels Article
dx.doi.org/10.1021/ef201353t | Energy Fuels 2012, 26, 854868 868
234
Centre of Combustion and Harmful Emission Control
Department of Chemical and
Biochemical Engineering
Technical University of Denmark
Sltofts Plads, Building 229
DK-2800 Kgs. Lyngby
Denmark
Phone: +45 4525 2800
Fax: +45 4525 4588
Web: www.chec.kt.dtu.dk
ISBN : 978-87-92481-66-5